You are on page 1of 47

COMENIUS UNIVERSITY IN BRATISLAVA

FACULTY OF MATHEMATICS, PHYSICS AND INFORMATICS

Fundamental group of some matrix groups


Bachelor thesis

May 2017 Kristna Blaskova


COMENIUS UNIVERSITY, BRATISLAVA
FACULTY OF MATHEMATICS, PHYSICS AND INFORMATICS

Fundamental group of some matrix groups


Bachelor thesis

Study program: Physics


Subject area: Physics
Number of subject area: 1160
Supervisor: Prof. RNDr. Pavol Zlatos, CSc.

Bratislava, May 2017 Kristna Blaskova


49068165

Univerzita Komenskho v Bratislave


Fakulta matematiky, fyziky a informatiky

ZADANIE ZVERENEJ PRCE

Meno a priezvisko tudenta: Kristna Blakov


tudijn program: fyzika (Jednoodborov tdium, bakalrsky I. st., denn
forma)
tudijn odbor: fyzika
Typ zverenej prce: bakalrska
Jazyk zverenej prce: anglick
Sekundrny jazyk: slovensk

Nzov: Fundamental group of some matrix groups


Fundamentlna grupa niektorch maticovch grp
Cie: Oboznmi sa s kontrukciou fundamentlnej grupy metrickho priestoru ako
jeho dleitho homotopickho invariantu a vypota fundamentlne grupy
niektorch vznamnch maticovch grp.

Vedci: prof. RNDr. Pavol Zlato, PhD.


Katedra: FMFI.KAGDM - Katedra algebry, geometrie a didaktiky matematiky
Vedci katedry: doc. RNDr. Pavel Chalmoviansk, PhD.
Spsob sprstupnenia elektronickej verzie prce:
bez obmedzenia
Dtum zadania: 17.04.2015

Dtum schvlenia: 21.07.2015 prof. RNDr. Jozef Masarik, DrSc.


garant tudijnho programu

tudent vedci prce


I would like to thank my supervisor Prof. RNDr. Pavol Zlatos, CSc.
for his never changing positive approach, patience and time he devoted
to me.
Abstract

BLASKOVA, Kristna, Fundamental group of some matrix groups [Bachelor


thesis], Comenius University in Bratislava, Faculty of Mathematics, Physics and
Informatics, Department of Algebra, Geometry and Didactics of mathematics;
Supervisor Prof. RNDr. Pavol Zlatos, CSc. Bratislava, 2017, 47p

The aim of this work is to calculate the fundamental group of some important
matrix groups. In the Basic knowledge we introduce the notions of metric and
topological spaces and their important properties such as path-connectedness,
connectedness, and simple-connectedness. These notions are necessary to talk
about paths, loops and their homotopy. Two paths are homotopic if and only if
they can be continuously deformed into each other while keeping their endpoints
fixed. Such paths form equivalence classes. It turns out that such homotopy
equivalence classes of loops (paths whose initial point is equal to the terminal
point) even form a group (fundamental group) endowed with the binary oper-
ation of path multiplication *- for example for a * b we first follow the loop a
twice the speed and then the loop b twice the speed. In the second chapter
The fundamental group, the fundamental group of topological groups is of our
special interest, because of its frequent appearance in physics. Here we show we
can introduce yet another definition of the fundamental group when it comes to
the topological groups. Finally, we calculate the fundamental group of SO(3). In
Conclusion, we provide a guide to Diracs scissors experiment, which illustrates
why the rotating objects can have only a one half spin.
key words: loops, homotopy, fundamental group
Abstrakt

BLASKOVA, Kristna, Fundamentalna grupa niektorych maticovych grup


[Bakalarska praca], Univerzita Komenskeho v Bratislave, Fakulta matematiky,
fyziky a informatiky, Katedra algebry, geometrie a didaktiky matematiky; Veduci
prace Prof. RNDr. Pavol Zlatos, CSc. Bratislava, 2017, 47p

Cielom tejto prace je spoctat fundamentalnu grupu niektorych vyznamnych


maticovych grup. V Zakladnych vedomostiach uvadzame pojmy metrickych a
topologickych priestorov, ako aj ich dolezitych vlastnost ako oblukova suvislost,
suvislost a jednoducha suvislost. Tieto pojmy su nevyhnutne na to, aby sme
mohli hovorit o cestach, sluckach a ich homotopii. Dve cesty su homotopicke
prave vtedy, ked sa daju spojito deformovat jedna na druhu, pricom fixujeme
ich koncove body. Taketo cesty formuju triedy ekvivalencie. Zistme, ze taketo
homotopicke triedy ekvivalencie sluciek (ciest, ktore maju rovnaky pociatocny a
koncovy bod) dokonca tvoria grupu (fundamentalnu grupu) vybavenu binarnou
operaciou skladania sluciek * - naprklad pre a * b najprv nasledujeme slucku a
dvojnasobnou rychlostou a potom slucku b taktiez dvojnasobnou rychlostou.
V druhej kapitole Fundamentalna grupa sa specialne zaujmame o fundamentalnu
grupu topologickej grupy, nakolko sa casto vyskytuje vo fyzike. Ukazeme, ze
mozeme zaviest este inu definciu fundamentalnej grupy v suvislosti s topo-
logickou grupou. Nakoniec spoctame fundamentalnu grupu SO(3). V Zavere
ponukneme navod na Diracov experiment s noznicami, ktory ilustruje, preco
mozu mat rotujuce objekty iba polocsleny spin.
key words: slucky, homotopia, fundamentalna grupa
Preface
Although this work on fundamental group is only a bachelor thesis, I strongly hope
it will be a source of knowledge for many alike people and students. Hundreds of
hours spent googling for better understanding of something that was once written into
one tiny sentence, that is mathematics in my view. This work being a compromise
between my very informal nature (and sometimes the lack of time, or laziness) and
mathematical exactness, is only a small window into algebraic topology and possible
solutions and proofs of fundamental groups of S 1 and SO(3). Working and searching
for information on my own, with the help of prof. Zlatos made me see that me,
and possibly many other students, get very passive during the studies. The only goal
becomes understanding of the proofs seen in the lectures and we forget that the beauty
is in the diversity of the ways one can come to the same result. Therefore this work
taught me not only about mathematics, but about the very basic approach to work as
well.
Contents
Introduction 9

1 Basic knowledge 10
1.1 Metric spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2 Topological spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3 Connected spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.4 Paths and loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.5 Homotopy of continuous mappings . . . . . . . . . . . . . . . . . . . . 18
1.6 Homotopy equivalence vs. Homeomorphism . . . . . . . . . . . . . . . 20

2 The fundamental group 23


2.1 Induced homomorphism . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.2 The fundamental group of circle . . . . . . . . . . . . . . . . . . . . . . 29
2.3 Fundamental group of topological groups . . . . . . . . . . . . . . . . . 36
2.4 The fundamental group of SO(3) . . . . . . . . . . . . . . . . . . . . . 40

Conclusion 45
Introduction
As the name (A First Course in Algebraic Topology) of the main reference book
indicates, this work is a very basic introductory to algebraic topology. Algebraic
topology provides a powerful tool when it comes to topological structure of spaces.
To every topological space we can associate a group, such that homeomorphic spaces
give rise to isomorphic groups. We are about to study one of such groups called
fundamental group. Throughout this thesis we will assume the knowledge of basic
group theory. Definitions and theorems in the first chapter can be found in [5], if not
stated differently. We will start by introducing topological spaces via the ideas based on
metric spaces. We will show it is enough to preserve the openness of set in the inverse
function to prove a function is continuous. To absolutely get rid off metric spaces, we
define the topology as a collection of subsets satisfying certain properties. Then the
original set together with this topology is called a topological space. This allows us
to move to nice spaces, i.e. connected spaces. Even if a space is not connected we
can divide it to set of connected spaces and approach each of them individually. Then
we introduce yet nicer spaces, they are called path-connected spaces. Together with
them we define loops, which when homotopic (continuously deformable into each other
keeping the base point fixed) form an equivalence class. These homotopy equivalence
classes form a group (fundamental group) endowed with the binary operation of path
multiplication *- for example for a * b we first follow the loop a twice the speed
and then the loop b twice the speed. Obviously, if the fundamental group is trivial,
there is only one equivalence class, the one consisting of loops contractible to a point
and hence there is no hole in this space. The simplest example of space with a hole is
circle, therefore we continue by calculating it. At this point we introduce topological
groups. Examples of these groups are Euclidean n-space Rn , the general linear group
GL(n, R), the orthogonal group O(n),... We further prove that the fundamental group
of topological group is abelian, moreover the set of all loops equipped with the point-
wise binary operation of topological group forms a group, too! We show that loops
equivalent to the identity form a normal subgroup and the respective quotient group
is not only equivalent as set to the fundamental group, but they binary operations
coincide. Then we move to calculation of the fundamental group of SO(3), we briefly
remind the reader the quaternion theory and introduce two theorems from covering
spaces to be able to calculate the fundamental group of SO(3). In the Conclusion we
illustrate that in physics a rotating object can have only a spin of a half using the
Diracs scissors experiment.

9
1 Basic knowledge

1.1 Metric spaces


We intuitively use the notion of distance since our childhood. At first we compare the
distance of objects the pen is closer than the pencil, later we start to recognise there
is something as units of length and we say that an object is 5 meter far away from us,
what means that every meter is equally distant and that it is 100 centimetres. But
102 centimetres would be a good meter too. Furthermore, we can say that the first
meter will be defined as 102 centimetres, the second one as 105 centimetres, the third
one as 108 cm and so on. So the unit of length (meter) is a special case of something
more general. Being inspired by what we experience in our everyday life, we might find
some conditions that are always true and independent of how we define the distance
in our previous example. The general thing is called metrics and it possesses three
conditions.

Definition 1.1 Let A be a set. A function d : A A R satisfying

d(a, b) = 0 only and only if a = b

d(a, b) = d(b, a)

d(a, b) + d(b, c) d(a, c) for all a, b, c A - triangle inequality

is called a metric on A. A set A with a particular metric on it is called a metric space


and is denoted by (A, d) or simply M .

The first condition says that if two objects are zero distance from each other, then
they are at the same place and vice versa.
The second one says that the distance is unique for both observers. If your friend
lives 2km far away, then he would say the same thing that is I live from you 2km far
away.
And finally, the third one, known as triangle inequality: if you go home from school
and pick up your friend on the way home, you either travel the same school-home
distance, or more.
Changing our way of thinking of distance from distance to distance function (metric)
we should reconsider the classic definition of continuity.

10
Definition 1.2 Let (A, dA ), (B, dB ) be metric spaces. A function f : A B is said
to be continuous at x A if and only if for all  > 0 there exists > 0 such that
d(f (x), f (y)) <  whenever d(x, y) < . The function is said to be continuous if it is
continuous at all points x A.

So the classic definition was nothing else than special case of metric, called Eu-
clidean metric. But do we really need to define a metric space to be able to decide
whether a function is continuous? Let us consider a set of all real numbers with Eu-
clidean metric on it. Given a certain point, for example x = 0, one can speak of all
points being within  > 0. Let  = 1, we consider the interval (1, 1). Now change to
 = 0.5. Clearly, these points approximate x to a greater degree of accuracy compared
to the first ones. Since this  can be arbitrary small, we will see that it is the openness
of sets of inverse images that matters, not the metric.

Definition 1.3 A subset U of a metric space (A, d) is said to be open if for all x U
there exists an  > 0 such that if y A and d(y, x) <  then y U .

Theorem 1.1 A function f : M1 M2 between two metric spaces is continuous if


and only if for all open sets U in M2 the set f 1 (U ) is open in M1 .
In other words, f is continuous if and only if inverse images of open sets are open.

Proof. Let d1 and d2 denote the metrics on M1 and M2 respectively. Suppose that f
is continuous and suppose that U is an open subset of M2 . Let x f 1 (U ) so that
f (x) U . Now, there exists  > 0 such that B (f (x)) f 1 (U ) since U is open. The
continuity of f assures that there is > 0 such that

d1 (x, y) < d2 (f (x), f (y)) < ,

or in other words f (B (x)) B (f (x)) U which means that B (x) f 1 (U ). Since


this is so for all x f 1 (U ) it follows that f 1 (U ) is an open subset of M1 .
Conversely let x M1 ; then for all  > 0 the set B (f (x)) is an open subset of
M2 so that f 1 (B (f (x))) is an open subset of M1 . But this means that since x
f 1 (B (f (x))) there is some > 0 with B (x) f 1 (B (f (x))), i.e. f (B (x))
B (f (x)). In other words there is a > 0 such that d2 (f (x), f (y)) <  whenever
d1 (x, y) < , i.e. f is continuous.

Note that we defined the open set with a help of metric. If we do not want it to be
just an alternative definition of continuity on metric spaces, we have to come up with
another definition of space - a topological space.

11
1.2 Topological spaces
Definition 1.4 Let X be a set and let U be a collection of subsets of X satisfying

U, X U

the intersection of two members of U is in U

the union of any number of members of U is in U

Such a collection U of subsets of X s called a topology on X. The set X together with


U is called a topological space and is denoted by (X, U )

Note that the second condition means that the intersection of finite members of U
is in U . Now why is that. Let us consider the open interval (1/n, 1/n). It specifies
all points that are 1/n-close to 0, but the intersection excludes everything.
Since we gave up our metric point of view, the continuity of functions between
two topological spaces will be itself a definition.

Definition 1.5 A function f : X Y between two topological spaces is said to be


continuous if for every open set U of Y the inverse image f 1 (U ) is open in X.

Theorem 1.2 Let X, Y and Z be topological spaces. If f : X Y and g : Y Z are


continuous functions then the composite h = f g : X Z is also continuous.

Proof. If U is open in Z then g 1 (U ) is open in Y and so f 1 (g 1 (U )) is open in X.


But (f g)1 (U ) = f 1 (g 1 (U )).

Definition 1.6 Let X and Y be topological spaces. We say that X and Y are home-
omorphic if there exist inverse continuous functions f : X Y and g : Y X (i.e.
gf = 1Y and f g = 1X ). We write X = Y and say that f and g are homeomorphisms
between X and Y .

An equivalent definition would be to require a function f : X Y to be bijective,


continuous and its inverse to be also continuous. Two topological spaces are consid-
ered to be equivalent if there is homeomorphism between them. It preserves all the
topological properties of a given space. For example connectedness, compactness and
metrizability. In general any property that can be expressed using open sets.

12
1.3 Connected spaces
Definition 1.7 A topological space X is connected if the only subsets of X which are
both open and closed are and X. A subset of X is connected if it is connected as a
space with the induced topology.

In other words a topological space is connected if it consists of one piece. How do


we connect the definition with these words? Of that speaks the following theorem.

Theorem 1.3 A space X is connected if and only if X is not the union of two disjoint
non-empty open subsets of X.

Proof. Let X be connected and suppose X = X1 X2 where X1 and X2 are disjoint


open subsets of X. Then X X1 = X2 and so X1 is both open and closed which
means that X1 = or X and X2 = X or respectively. In either case X is not the
union of two disjoint non-empty open subsets of X. Conversely, suppose that X is
not the union of two disjoint non-empty open subsets of X and let U X. If U is
both open and closed then X U is both open and closed. But since X is then the
disjoint union of the open sets U and X U one of these must be empty, i.e. U =
or U = X.

The connectedness is a consequence of the topology induced on the space, not a


property of the space itself. Let us consider following example: every interval I R
with the usual topology on it is connected, but not connected with the discrete topology
on it.

Proof. Suppose I = A B and A B = , A and B are both non-empty and open


in the subspace-topology of I R. Choose a A and b B and suppose a < b.
Let s := inf{x B | a < x}. Then in every neighbourhood of s there are points of
B (because of the definition of the infimum), but also of A, then if not s = a, then
a < s and the open interval (a, s) lies entirely in A. And so s cannot be an inner
point of A nor B, but this is a contradiction to the property that both A and B be
open and s A B. Now consider the discrete topology. Since in this topology all of
the one point subsets are open and closed, it is in contradiction with the definition of
connectedness and hence disconnected.

13
Theorem 1.4 The Image of a connected space under a continuous mapping is con-
nected.

Proof. Suppose that X is connected and f : X Y is a continuous surjective map.


If U is open and closed in Y then f 1 (U ) is open and closed in X which means that
f 1 (U ) = or X and U = or Y . Thus Y is connected.

Theorem 1.5 Suppose that {Yj ; j J} is a collection of connected subsets of a space


X. If Yj 6= then Y = Yj is connected.
jJ jJ

Proof. Suppose that U is a non-empty open and closed subset of Y . Then U Yi 6=


for some i J and U Yi is both open and closed in Yi . But Yi is connected so
U Yi = Yi and hence Yi U . The set Yi intersects every other Yj , j J and so U
intersects every Yj , j J. By repeating the argument we deduce that Yj U for all
j J and hence U = Y .

Theorem 1.6 Let X and Y be topological spaces. Then X and Y are connected if and
only if X Y is connected.

Proof. Suppose that X and Y are connected. Since X = X {y} and Y = {x} Y
for all x X, y Y we see that X
= X {y} and Y
= {x} Y are connected. Now
(X {y} {x} Y ) = {(x, y)} 6= and so (X {y} {x} Y ) is connected by
Theorem 1.5. We may write X Y = ((X {y} {x} Y )) 6= we deduce that
X Y is connected. Conversely, suppose that X Y is connected. That X and Y are
connected follows from 1.4 and the fact that X : X Y X and Y : X Y Y
are continuous surjective maps.

14
1.4 Paths and loops
Definition 1.8 A continuous mapping f : [0, 1] X is called a path in X. The point
f (0) is called the initial point and f (1) is called the final or terminal point. We say
that f joins f (0) and f (1), and that f is a path from f (0) to f (1).
The image is called a curve in X.

In fact it could be a continuous mapping from any interval [a, b], a, b R, a 6= b,


and it is only a matter of preference that we use its re-parametrization to [0, 1].
If f is a path in X and f is defined by f(t) = f (1 t) then f is also a path in X but
in an opposite direction.
If f and g are two paths in X for which the final point of f coincides with the
initial point of g then the function f g : [0, 1] X defined by
(
f (2t) if 0 t 1/2
(f g)(t) =
g(2t 1) if 1/2 t 1

is a path in X and it follows from the next lemma.

Lemma 1. Let W , X be topological spaces and suppose that W = A B with A, B


both closed subsets of W . If f : A X and g : B X are continuous functions such
that f (w) = g(w) for all w A B then h : W X defined by
(
f (w) if w A
h(w) =
g(w) if w B

is a continuous function.

Proof. Note that h is well defined. Suppose that C is a closed subset of X, then

h1 (C) = h1 (C) (A B)
= (h1 (C) A) (h1 (C) B)
= f 1 (C) g 1 (C)

Since f is continuous, f 1 (C) is closed in A and hence in W since A is closed in W .


Similarly g 1 (C) is closed in W . Hence h1 (C) is closed in W and h is continuous.

Definition 1.9 A space X is said to be path connected if given any two points x0 , x1
in X there is a path in X from x0 to x1 .

To verify the path connectedness when needed the following theorems will help us.

15
Theorem 1.7 The image of a path connected space under a continuous mapping is
path connected.

Proof. Suppose that X is path connected and g : X Y is a continuous surjective


map. If a, b are two points of Y then there are two points a0 , b0 in X with g(a0 ) = a
and g(b0 ) = b. Since X is path connected there is a path f from a0 to b0 . But then gf
is a path from a to b which shows that Y is path connected.

Theorem 1.8 Every path connected space is connected. Not every connected space is
path connected.

Proof. Suppose that X is a path connected space. We shall prove that X is connected.
To this end let X = U V with U , V open and non-empty. Since X is path connected
and U , V are non-empty there is a path f : [0, 1] X with f (0) U and f (1) V .
Since [0, 1] is connected so is f ([0, 1]) and so U f ([0, 1]), V f ([0, 1]) cannot be
disjoint. Thus neither can U and V be disjoint and so X is connected.
To show that not all connected spaces are path connected we shall give an example
known as the flea and comb. Consider the subset X C where X = A B with

A = {i} (the f lea)


B = [0, 1] {(1/n) + yi; n N, 0 y 1} (the comb)

We claim that X is connected but not path connected. To prove that X is connected
first observe that B is path connected (use Theorem 1.5 as an analogy to path con-
nectedness on Bn = [0, 1] {(1/n) + yi, 0 y 1} for n N and then B = Bn
nN
and so B is connected. Suppose that U is an open and closed subset of X. We may
suppose that A U (otherwise the complement of U is an open and closed subset
of X that contains A). Since U is open and i U there is an  > 0 such that
{x; |i x| < } X U . There is some integer n such that (1/n) + i U ; in particular
U B 6= But B is connected and U B is a non-empty open and closed subset of
B; thus U B = B, i.e. B U . But X = A B and A U ; thus U = X and so X
is connected.
To prove that X is not path connected we shall show that the only path in X that
begins at i X is the constant path. Let f be the path in X that begins at i X.
Since i is closed in X, f 1 (i) is closed in [0, 1], furthermore f 1 (i) 6= since 0 f 1 (i).
Let U be the open subset of X given by U = X {z C; |z i| 21 }. If t0 f 1 (i)
then since f is continuous there is an  > 0 such that f (t) U whenever |t to | < .
We claim that f ((t0 , t0 + ) [0, 1]) = i. To see this suppose that |t1 t0 | <  and

16
f (t1 ) B. Since U B is a union of disjoint intervals, the interval containing f (t1 ) is
both open and closed in U (open because U is open, and closed because the interval
is of the form {(1/n) + yi, 0 y 1} U for some n N . But this contradicts the
fact that f ((t0 , t0 + ) [0, 1]) is connected. Hence

f ((t0 , t0 + ) [0, 1]) f 1 (i)

We have just shown that if t0 f 1 (i) then

f ((t0 , t0 + ) [0, 1]) f 1 (i),

which means that f 1 (i) is open. Since f 1 (i) is also closed and [0, 1] is connected it
follows that f 1 (i) = [0, 1]. There is therefore no path between i X and any point
B X; thus X is not path connected as was claimed.

17
1.5 Homotopy of continuous mappings
As it was natural to talk about equivalent spaces via homeomorphism, we also feel
that there are some paths which are more equivalent to each other than to others.
Later we will see that this equivalence of paths describes two spaces to be the same
shape in much broader way than homeomorphism. We will use an example from [3].
Consider the alphabet letters as on the Figure 1. These letters can be written either

Figure 1: Alphabet letters in form of curves [3]

as unions of finitely many straight and curved line segments, or in thickened forms
that are compact subsurfaces of the plane bounded by simple closed curves. In both
cases the thinner letters are subspaces of the thicker ones and can be continuously
transformed to each other. One of the ways to do it is to decompose the letter X into
line segments connecting each point on its outer boundary to a unique point of the
thin subletter X as on the Figure 1. Then we slide each point along the line segment
that contains it. Points in X do not move. We can think of this process as it would
be happening during a time interval 0 t 1. Then it defines a family of functions
ft : X X parametrized by t I = [0, 1], where ft (x) is the point where x X has
moved at time t.
Now we would like ft (x) to depend continuously on both t and x. This will be done
by moving at constant speed so as to reach its image point at time t = 1, while
points x X are stationary. This example is one of the special cases of the following
definition.

Definition 1.10 Two continuous maps f0 , f1 : X Y are said to be homotopic if


there is a continuous map F : X I Y such that F (x, 0) = f0 (x) and F (x, 1) =
f1 (x).
The map F is called a homotopy between f0 and f1 . We write f0 ' f1
or F : f0 ' f1 . For each t [0, 1], we denote F (x, t) by ft (x), whence ft : X Y is
a continuous map.

18
Definition 1.11 Suppose that A is a subset of X and that f0 , f1 are two continuous
maps from X to Y . We say that f0 and f1 are homotopic relative to A if there is
a homotopy F : X I Y between f0 and f1 such that F (a, t) does not depend on t
for a A; in other words F (a, t) = f0 (a) for all a A and all t I.
We write f0 ' f1 rel(A).

Lemma 2. The relation f0 ' f1 rel(A) on the set of continuous maps from X to Y
is an equivalence relation.

Proof. The relation is reflexive because F (x, t) = f (x) is a homotopy relative to A


between f and f itself. It is symmetric because if F : f ' g rel(A)
then G : g ' f rel(A) where G is given by G(x, t) = F (x, 1 t). Finally, the relation
is transitive because if F : f ' g and G : g ' h then H : f ' h where H is given by:
(
f (x, 2t) if 0 t 1/2
H(x, t) =
g(x, 2t 1) if 1/2 t 1

Lemma 1 shows that H is continuous.

Definition 1.12 We denote the equivalence class of f as [f ].

Definition 1.13 Two spaces X and Y are of the same homotopy type if there exist
continuous maps f : X Y , g : Y X such that

gf 'I :X X

f g 'I :Y Y

The maps f and g are then called homotopy equivalences. We also say that X and Y
are homotopy equivalent.

19
1.6 Homotopy equivalence vs. Homeomorphism
Often the question of difference between homotopy and homeomorphism arises. The
condition of homeomorphism is stronger than homotopy. Recall the definition of home-
omorphism (Definition 1.6). In homeomorphism the function composition f g and
g f must result into identity function. As seen in the Figure 2 for homotopy it does
not have to be true necessary. For example let f : X Y map three of the prongs of

Figure 2: Homotopy equivalent spaces

the letter X on to the letter Y so that the upper prongs map identically and one of
the lower prong just moves so that it creates Y . Let it map the fourth prong to the
point at the centre. Let g : Y X map the Y into those three prongs of the X. The
letter X and Y are not homeomorphic, but they are homotopic [1].

Proof. Obviously f and g are both continuous. f is a surjection, but not injective,
while g is an injection, but not surjective. Therefore they are not homeomorphic. Now
f g and g f are easily seen to be homotopic to the identity [1].

Definition 1.14 Two paths f , g in X are said to be equivalent if f and g are homo-
topic relative to {0, 1}. We write f g.

In other words the paths f0 , f1 in X are equivalent if there is a continuous map


F : I I X such that

F (t, 0) = f0 (t) and F (t, 1) = f1 (t) for t I


F (0, s) = f0 (0) and F (1, s) = f0 (1) for s I

20
Figure 3: Equivalent paths

Lemma 3. Suppose that f , g, h are three paths in X with f (1) = g(0) and g(1) = h(0).
Then (f g) h f (g h).

Proof. Note first that



1


f (4t) if 0 t 4
((f g) h)(t) = 1 1
g(4t 1) if 4
t 2

h(2t 1) if 1
t1

2

and
1


f (2t) if 0 t 2
(f (g h))(t) = 1 3
g(4t 2) if 2
t 4

h(4t 3) if 3
t1

4

To construct a homotopy, for a given value of s, we use f in the interval [0, s+1
4
] , g in
the interval [ s+1
4
, s+2
4
] and h in the interval [ s+2
4
, 1]. The function we are looking for is
just a linear transformation of these intervals to the interval [0, 1]. So we get

4t


f ( 1+s ) if 0 t s+1 4
s+1
F (t, s) = g(4t s 1) if 4
t s+2
4

h( 4ts2 )

if s+2
t 1
2s 4

The function F is continuous and


F (t, 0) = ((f g) h)(t) and F (t, 1) = (f (g h))(t) for t I
F (0, s) = ((f g) h)(0) and F (1, s) = (f (g h))(1) for s I
so that F provides the required homotopy.
If x X then we have defined x : I X as the constant path, i.e. x (t) = x. The
equivalence path of the constant path behaves as a (left or right) identity, i.e.

[ex ][f ] = [f ] = [f ][ey ]

if f is a path that begins at x and ends at y. This is proved in the next result.

21
Lemma 4. If f is a path in X that begins at x and ends at y then x f f and
f y f .

Proof. We shall only prove that x f f , the proof f y f is similar.


Define I I X by
(
1s
x if 0 t 2
F (t, s) =
f ( 2t1+s
1+s
) if 1s
2
t1
Then F (t, 0) = x f and F is a homotopy relative to {0, 1}.
Finally we would like inverses to paths (up to equivalence of paths). To this end
recall that if f is a path then f is the path defined by f(t) = f (1 t). Note that f g
if and only if f g. The next result will show that the equivalence class of f acts as
an inverse for the equivalence class of f , i.e.

[f ][f] = [x ], [f][f ] = [y ]

for a path f beginning at x and ending at y.

Lemma 5. Let f be a path in X that begins at x and ends at y. Then f f x and


f f y .

Proof. We shall only prove that f f x . The path f f is given by


(
f (2t) if 0 t 12
(f f)(t) =
f (2 2t) if 21 t 1
It represents a path in which we travel along f for the first half of our time interval
and then in the opposite direction along f for the second half. To make sure that we
get from x to y and back to x we travel at speed 2 (i.e. twice normal speed). If we
now vary the speed proportionally to (1 s) for s I then for each s we get a path
that starts at x, goes to f (2(1 s)) and then returns to x. For s = 0 we get f f and
for s = 1 we get x . Therefore define F : I I X by
(
f (2t(1 s)) if 0 t 12
F (t, s) =
f ( 22t
1s
) if 21 t 1
F is obviously continuous and

F (t, 0) = (f f)(t) F (t, 1) = f (0) = x (t)


F (0, s) = f (0) = (f f)(0) F (1, s) = f (0) = (f f)(1)

so that f f x .

22
2 The fundamental group
Let us start with an example from [3]. Consider two circles A and B in R3 as shown
in the Figure 4. From our physical world we are used to that no matter what we do,

Figure 4: Two inseparable circles. [3]

whether we push, twist or pull the circles, we cannot separate them. We can even
imagine them to be made of rubber and stretch them into different shapes, but this
does not help us with separating them. The fundamental group is here to move us
from the intuition into mathematically rigorous theory. Instead of having A link with
B just once, we can make it link several times as in the Figure 5.

Figure 5: Two oriented circles linked several times. [3]

As a further variation, we can let B link A in some direction positive or negative


times, say positive when B comes forward through A and negative for the reverse
direction. Thus for each nonzero integer n we have an oriented circle Bn linking A n
times, where by circle we mean a curve homeomorphic to a circle. To complete the
scheme, we could let B0 be a circle not linked to A at all.
These integers even form a group under addition. Can be the group operation
visualised geometrically with some additive operation on the oriented circles? We can
think of the oriented circle B as of a path traversed in time t, starting and ending at
the same point x, as it is Definition 2.1.

23
Definition 2.1 A path is said to be closed or equivalently called a loop if f (0) = f (1).
If f (0) = f (1) = x then we say that f is based at x.

Two different loops starting and ending at the same point x can be added to form
a new loop B + B 0 , which first travels around B and then around B 0 . For example, as
seen in the Figure 6, let B1 and B10 be paths in the positive direction, then their sum
B1 + B10 can be continuously deformed into B2 . Similarly, B1 + B1 can be deformed
to the loop B0 , unlinked from A. In general, a loop Bn + Bm can be deformed to a
loop linked n + m times to A. The important difference between our intuition in the
physical world and the world of loops is that by summing loops we cross the base point
several times. So if one is thinking about the loop as about a string, one has to give

Figure 6: Deformable loops. [3]

the string the magical power to cross the same point more than once.
Next we consider a little bit more complicated example. It involves three circles
known as the Borromean rings.

Figure 7: Deforming the Borromean rings. [3]

The interesting thing here is that whenever we remove one of the circles the other
two are not linked. The same way as before let us choose the circle C as a loop in the
complement of the other two. Again we try to separate those rings by pulling them

24
apart and by allowing C to be continuously deformed. Now if we follow the path which
base point is defined by dot in the Figure 7 and the direction by the arrow, we can
notice that if we pass in sequence: (1) forward through A, (2) forward through B, (3)
backward through A, and (4) backward through B and forward and backward with
respective integers, their sum cancels. This reflects the fact that C is not linked with
A or B individually.

Figure 8: Visualising one of the rings as several paths. [3]

To describe more accurately how C links with A and B together, we imagine C as


sum of four paths as seen in the Figure 8. Taking into account the directions in which
these segments of C pass through A and B, we may deform C to the sum a+bab of
four loops as in the figure. We write the third and fourth loops as the negatives of the
first two since they can be deformed to the first two, but with the opposite orientations,
and as we saw in the preceding example, the sum of two oppositely oriented loops is
deformable to a trivial loop, not linked with anything. But what if it is nonabelian?
Then it does not automatically sums to zero.

Figure 9: Visualising one of the rings as several paths. [3]

Now let us take this example a little bit further and see the Figure 9. The circles A
and B are now linked. The circle C can then be deformed into the position shown at
the right, where it again represents the composite loop a b a b, where we used the

25
usual composition notation instead of previously used sum. Notice that now C can be
completely unlinked from A and B, so here should be the product a b a b trivial.
The fundamental group of a space X will be defined so that its elements are loops in
X starting and ending at a fixed basepoint x X, but two such loops are regarded as
determining the same element of the fundamental group if one loop can be continuously
deformed to the other within the space X. We denote the set of equivalence classes
of closed paths based at x X by (X, x) and in the next theorem we prove it is a
group.

Theorem 2.1 (X, x) is a group equipped with binary operation of path composition
*.

Proof. We have already verified the group conditions. Closure by Lemma 1, asso-
ciativity by Lemma 3, identity element by Lemma 4 and inverse element by Lemma
5.

It is very natural to ask to what extent depends the fundamental group on the
choice of base point. Since (X, x) involves only the path-component of X containing
x, it is clear that we can hope to find a relation between (X, x0 ) and (X, x1 ) for
two basepoints x0 and x1 only if x0 and x1 lie in the same path-component of X. So
let h : I X be a path from x0 to x1 , with the inverse path h(s) = h(1 s) from
x1 back to x0 . We can then associate to each loop f based at x1 the loop h f h
based at x0 . The order of path composition does not matter here, since (h f ) h and
h (f h) are homotopic and we are only interested in homotopy classes.

Figure 10: The change of basepoint. [3]

Theorem 2.2 Let x0 , x1 X. If there is a path in X from x0 to x1 then the groups


(X, x0 ), (X, x1 ) are isomorphic.[5]

Proof. Let us recall the definition of group isomorphism first. Two groups are isomor-
phic if and only if there exists a bijective function h : G H such that for all
a, b G it holds that h([a][b]) = h[a]h[b].

26
Let h be a path from x0 to x1 . If f is a closed path based at x1 then (h f ) h
is a closed path based at x0 . We therefore define

uh : (X, x1 ) (X, x0 )

by
uh [f ] = [h f h]

This is a homomorphism of groups because

uh ([g][f ]) = uh [g f ]
= [h g f h]
= [h g h h f h]
= [h g h][h f h]
= uh [g]uh [f ]

Using the path h from x0 to x1 we can define

uh : (X, x0 ) (X, x1 )

by
uh = [h f h]

A simple check now shows that uh uh [g] = [g] and uh uh [f ] = [f ] so that uh is bijective
and hence is an isomorphism.

So if X is a path connected space then (X, x0 ) and (X, x1 ) are isomorphic fun-
damental groups, for any pair of points x0 , x1 X.

2.1 Induced homomorphism


Exactly as we have questioned the behaviour of fundamental group under the change of
base point, we can go further by asking about what happens if we affect it by mapping.
Let : X Y be a continuous map; the following three statements are true [5]:

If f , g are paths in X then f ,g are paths in Y .

If f g then f g.

If f is a closed path in X based at x X then f is a closed path in Y based


at (x).

27
Thus if [f ] (X, x), then [f ] is a well defined element of (Y, (x)). We therefore
define
: (X, x) (Y, (x))

by
[f ] = [f ]

Lemma 6. is a homomorphism of groups and we call it induced homomorphism


[5].

Proof. ([f ][g]) = [f g] = [(f g)] = [f g] = [f ][g] = [f ] [g]

Theorem 2.3 Suppose : X Y and : Y Z are continuous maps, then


( ) = [5].

Proof. ( ) [f ] = [ f ] = [ f ] = [f ]

Theorem 2.4 If id : X X is the identity map, then id is the identity homomor-


phism on (X, x) [5].

28
2.2 The fundamental group of circle
Now it is the time to connect theory and calculation. Here we calculate the most
basic example of nontrivial fundamental group: the fundamental group of a circle.
Theorems, proofs and lemmas in this section are inspired by Hatcher [3].

Figure 11: Sketch of the loops in circle.

It is very easy to come up with the solution intuitively. The only kind of loop is
the one around the circle. The composition of this loop with itself gives us a loop that
runs around the circle two times. However, this composed loop is not homotopic to
the loop which runs around the circle only once. Exactly the same way we can create
loops running n times around the circle not being homotopic to any other loop than to
itself. Moreover we can play with the direction of the loops, let us say positive when
clockwise and negative when anticlockwise. All of these are homotopy classes of the
fundamental group of circle. Therefore we can say that (S 1 )
= Z.
In this proof we will use

n (s) = (cos(2ns), sin(2ns))

for s [0, 1], n Z to describe the loop of a unit circle S 1 , where n represents the
number of times this path winds around the circle. The main idea of this proof is to
compare paths in S 1 with paths in R. We are going to do this by imagining R to be
a helix in R3 as seen in the Figure 12. By projecting the helix onto a plane, we get a
circle. We define this projection mapping p : R S 1 by

p(s) = (cos(2s), sin(2s)).

Now let us be more specific about what paths are we about to compare. In R it will
en : [0, 1] R,
be the paths defined by


en (s) = ns.

29
2

Figure 12: Projection of R onto S 1 .

This is simply a straight line from 0 to n. In the helix it starts at 0 and winds around
|n| times as seen in the Figure 12 with a special case of n = 2. Next, define n : I S 1
to be the loop
n (s) = (cos(2ns), sin(2ns)).
Visually, n (I) starts at (1, 0), wraps |n| times around S 1 (clockwise if n < 0 and
counterclockwise if n > 0) and ends back at (1, 0). Here we pause to observe that
n = p
en which allows us to write down the following commuting diagram:

Figure 13: The change of basepoint. [3]

We call
en a lift of n . In the proof we are about to use the following two lemmas
in connection with lift.

e0 p1 (x0 ), there
Lemma 7. For each path f : I S 1 , with f (0) = x0 and for each x
is a unique lift fe : I R such that fe(0) = x
e0 .

30
Lemma 8. For each homotopy ft : I S 1 of paths starting at x0 and for each
e0 p1 (x0 ), there is a unique lift fet : I R such that fet (0) = x
x e0 .
Both lemmas are just special cases of the following, more general theorem.
Theorem 2.5 Given any space Y , a map F : Y I S 1 , and a map Fe : Y 0 R,
which lifts F |Y {0} , there exists a unique map G : Y I R which lifts F and is an
extension of Fe.

Figure 14: Notice that Lemma 7 and Lemma 8 are the special cases when Y = {y0 }
is a single point and Y = I is the unit interval, respectively.

Figure 15: An example of open cover of S 1

Proof. We begin with an observation: There is an open cover {U } of S 1 such that


for each , the set p1 (U ) is a disjoint union of open sets each of which is mapped
homeomorphically to U by p.

31
For example, let {U } = {U1 , U2 } be an open cover of S 1 , such that
U1 = S 1 \ {(1, 0)} and U2 = S 1 \ {(1, 0)} as visualized in the Figure 15. Then
p1 (U1 ) = {(n, n + 1)}nZ is a disjoint collection and p(n, n + 1) = U1 for all n.
1 n1 n+1 n1 n+1
Similarly p (U2 ) = { 2
, }nZ is disjoint and p(
2)
,
2
) = U2 for all n.
2)

Figure 16

Suppose F : Y I S 1 is any map and let Fe : Y {0} R be any map which


lifts F |Y {0} . Let (y0 , t) Y I be any point and observe that it has a neighbourhood
Nt (at , bt ) Y I such that (Nt (at , bt )) U S 1 for some . (This last bit
holds simply because {U } is a cover for S 1 .)
By compactness of {y0 } I (since its homeomorphic to I), finitely many such
products Nt (at , bt ) cover y0 I. Thus we can choose a single neighbourhood N of
y0 and a partition 0 = t0 < t1 <...< stm = I0 of I so that for each i, F (N [ti , ti+1 ])
is contained in U for some , which we will now denote by Ui . Our present goal is to
construct a lift G : N I R. Assume by induction on i that G has already been
constructed on N [0, ti ]. (Notice! This is true for i = 0 since N [0, 0] is a subset of
Y {0} and by assumption we already have the map Fe : Y {0} R.) Now since
F (N [ti , ti+1 ]) Ui we know by our opening observation that p1 (Ui ) is a disjoint
union of open sets in R, each of which is mapped homeomorphically onto Ui . Hence
ei p1 (Ui ) which contains the point G(y0 , ti ) (this point exists by the
there exists a U
induction hypothesis) since (y0 , ti ) sits inside N [0, ti ], and the latter maps into Ui
by F . Now its possible that G(N {ti }) (the lift of the top slice of N [0, ti ]) isnt

32
fully contained within U ei but we can just assume it is, since we can replace N ti with
1
the intersection (N ti ) G N {ti } (Ui ). Finally we may define G on N [ti , ti+1 ] by
 1
p ui F [ti ,tt+1 ] .

This is well-defined since p is a homeomorphism, so we can repeat this inductive
ui
step finitely many times to obtain the desired lift G : N I R for some neighborhood
N of y0 . And since y0 Y was arbitrary, we have our lift G on all of Y I.

Now we are finally ready to move to the proof that the fundamental group of the
circle is Z.

Theorem 2.6 The map : Z (S 1 ) given by n 7 [n ], is an isomorphism. In


other words (S 1 )
= Z.

Proof. We are going to divide this proof into 3 parts that will hopefully help the reader
not to get lost in this quite lengthy proof.

1. is a group homomorphism.

2. is surjective.

3. is one on one.

1. We would like to begin with proving is a homomorphism i.e. m+n = m n .


We define a translation map m : R R by x 7 x + m for m Z. Notice that in
our helix model of R, this is a shift up or down by |m|. This observation lets us
conclude that
m m n
=em+n

by some homotopy F : I I R. On the right


em+n is simply the line (path) from 0
to m + n in R. On the left m m n is the product of the path from 0 to m with
the path from m to m + n. So when we project each of these paths onto S 1 , we see
that the resulting paths are also homotopic:

p (m m n )
=p
em+n

with homotopy G = p F .

33
Now we have enough information to write the following equalities:

m n = [m ] [n ] = [p
em ] [p
en ]
= [p
em ] [p (m
en )]
= [(p
em ) (p m
en )]
= [p (e
m m
en )]
= [p
em+n ]
= [m+n ]
= m+n

So is indeed a homomorphism.
2. We continue by proving is surjective. To prove this, we will use the Lemma 7 (see
bellow). This lemma tells us that not only path in the helix (R) will project down to
a path in S 1 but the converse is true, too. That is, given a path f downstairs in S 1
that starts at x0 , and given a lift x
e0 of x0 , there is a path fe upstairs in R starting at
e0 that projects down to f . Notice the set p1 (x0 ) is infinite! So the lemma says that
x
e0 p1 (x0 ) (we call this a point in the fiber above x0 ), then there is
if you fix an x
a map fe : I R which takes 0 to that x e0 .
Let [f ] (S 1 )[f ] be the path-homotopy class of some loop in S 1 and pick a
representative f : I S 1 with base point (1, 0) = f (0) = f (1). We wish to show that
there is an n Z such that (n) = [f ], i.e. such that f
= n for some n.
To this end, begin by observing that 0 p1 (1, 0) = Z. Hence by Lemma 7, there is
a unique liftfe : I R such that fe(0) = 0. Now since

(p fe)(1) = f (1) = (1, 0) S 1

we see that fe(1) must be an integer. In other words, fe is a path from 0 to n for some
integer n. Thus fe = en for some homotopy F (they are homotopic and not equal
en (s) = ns for s [0, 1]) and so
because fe isnt necessarily the straight line

f = p fe
=p
en = n

With this we conclude [f ] = [n ] = n and so is surjective.


3. To prove is one on one we will use Lemma 8.
Suppose (m) = (n) so that m = n , for some m, n Z. We wish to show m = n.
Let ft : I S 1 be the homotopy from m to n with f0 = m and f1 = n and notice
that ft (0) = (1, 0) S 1 . This follows since (1.0) is the common starting point of m
and n . Also note that 0 p1 (1, 0) = Z.

34
By Lemma 8, there exists a unique lift fet : I R such that fet (0) = 0 R. More-
over, since the lift fet (0) is unique, we must have fe0 =
em and fe1 =
en .
Lemma 8 tells us that the given lift fet of ft is unique. So if fe0 is any other lift
off0 = m , it must be that fe0 =
em . A similar statement holds for the two liftsfe1 and

en of n = f1 .
Sincefet is a path homotopy, fet (1) (the shared endpoint of the loops that fet is homotopy-
ing) must be constant for all t. In particular, we must have fet (1) =
em (1) =

en (1) =the common end point of


em and
en . But this common end point must
be an integer, and it must be both m and n. Hence m = n.

35
2.3 Fundamental group of topological groups
Very interesting to us are fundamentals groups of topological groups, as these groups
are very likely to appear in physics. For this purpose we define the topological group
as following.

Definition 2.2 A topological group G is a group that is also a topological space in


which the maps : G G G and : G G defined by (g1 , g2 ) = g1 g2 and
(g) = g 1 are continuous.

We mention some typical examples, so reader becomes easily aware of the topolog-
ical groups widely used in physics.

EXAMPLES [4]

1. The additive group R or C with the usual metric topology. The group operation
is addition and the inverse operation is negation

2. The multiplicative group of positive real numbers

3. The additive group Rn

4. Any subgroup of topological group with the relative topology. Thus, for example,
the circle {z C; |z| = 1} is a subgroup of the multiplicative group C \ {0}.

5. The general linear group GL(N, C) of all non-singular N-by-N complex matrices,
with matrix multiplication as a group operation. The topology is the relative
2
topology from CN . Each entry in a matrix is a polynomial in the 2N 2 entries of
the two matrices being multiplied and is therefore continuous; thus matrix mul-
tiplication is continuous. Inversion is defined on the set where the determinant
polynomial is not zero and is given, according to Cramers rule, in each entry
by the quotient of a polynomial function and the determinant function; thus in-
version is continuous. By (4) the general linear group GL(N, R) is a topological
group.

To investigate the behaviour of topological group, we will question its behaviour with
loops. In the next theorem we would like to show, that the set of loops forms group
even under topological group multiplication.

Theorem 2.7 Let G be a topological group with operation and identity element e. Let
(G, e) denote the set of all loops in G based at e. If f , g (G, e), then (G, e) is a
group endowed with the point-wise binary operation defined as (f g)(s) = f (s) g(s).

36
Proof. We check that f g is a loop. Its endpoints (f g)(0) and (f g)(1) match because
they do for the component loops. It is a continuous function because multiplication is
continuous by definition of topological group.
Now we verify the group properties:

1. Associativity of composition follows from associativity of multiplication in G.

2. The identity element is the constant loop g(s) = e. Indeed, (f g)(s) = f (s)
g(s) = f (s) e = f (s)

3. The inverse element for the loop f (s) is the loop h(s) = (f (s))1 . Indeed,
(f h)(s) = f (s) h(s) = f (s) (f (s))1 = e = g(s) as required.

Does this group contain any nontrivial subgroup? If yes, the most basic guess would
be a subgroup containing loops homotopic to the constant path, since every subgroup
must contain the identity element and from the nature of the binary operation on
topological groups (its continuity), we would expect some kind of connectedness, very
similar to homotopy, happening here. It turns out it is even a normal subgroup of the
group (G, e).

Theorem 2.8 The connected component of the identity (G) = {f G; f e} is a


normal subgroup of (G) [8].

Proof. That e (G) is trivial. We begin by showing that if a (G) (from now on
just ), then its inverse a1 is in , too. Let f : [0, 1] G be the path homotopy from
a to e. Since the inverse element function is on G, f 1 : [0, 1] G, f 1 (t) = f (t)1 is
a path from a1 to e.
Now we want to show that if a, b , then a b is in , too. Let f , g : [0, 1] G
be paths such that f is from a to e, g is from b to e. Since multiplication on G is
continuous, the path f g : [0, 1] G from a b to e is continuous and hence a b .
The last thing to show is the subgroup normality. Let a and x G, we want
to show that x a x1 . Let f : [0, 1] G, f (0) = a and f (1) = e. Then
h(t) = x f (t) x1 is a path h : [0, 1] G from x a x1 to e = x e x1 and hence
is a normal subgroup of (G).

Now let us try to compare (X, x) and (G)/(G) as sets. Recall that the funda-
mental group (X, x) is nothing else than this set modulo homotopy

{f : [0, 1] X : f (0) = x = f (1)}/

37
and when applied to topological group, it looks like this

(G, e)/

The next theorem will convince us that

(G, e)/ = (G, e)/(G, e)

i.e. as sets they are equal.

Theorem 2.9 Let f , g (G), then f g if and only if f g 1 (G). i.e.


f g 1 e [8].

Proof. Let f g. That means there exists homotopy H(s, t) such that H(0, t) = f (t)
and H(1, t) = g(t). We want to show that f g 1 e i.e. there exists homotopy G(s, t)
such that G(0, t) = f g 1 and G(1, t) = e. We can define G(s, t) and H(s, t) to be in
the following relation
G(s, t) = H(s, t) g 1 (t)

Now it is easy to verify that G(0, t) = H(0, t) g 1 (t) = f (t) g 1 (t) and
G(1, t) = H(1, t) g 1 (t) = g(t) g 1 (t) = e and since is a continuous operation,
f g 1 e.
Now we are about to prove that if f g 1 e then f g. Similary we define the
homotopy the following way

H(s, t) = G(s, t) g(t)

Again G(0, t) = f g 1 and G(1, t) = e. We check H(0, t) = G(0, t)g(t) = f g 1 g(t) =


f (t) and H(1, t) = G(1, t) g(t) = e g(t) = g(t). Hence f g.

Theorem 2.10 The binary operation of fundamental group coincides with the point-
wise operation of topological group .

Proof. The fundamental group element is an equivalence class, therefore we need to


show that the resulting element of both binary operations belongs to same equivalence
class. In other words
f g f g. (2.1)

That means (
f f g if 0 t 1/2
g f g if 1/2 t 1

38
so we can rewrite equation 2.1 to

f ef g (2.2)

and
eg f g (2.3)

This leads us to the idea of using a trick. If we re-parametrize f f e and g e g


on the right side of equation 2.1, we actually get what we want and that is

[f ] [g] = [f e] [e g] = [f e] [e g] = [f ] [g] (2.4)

Writing down the explicit homotopy:

H(t, s) = P (t, s) Q(t, s)

where (
2t 1+s
f ( 1+s ) if 0 t 2
P (t, s) = 1+s
e if 2
t1
and (
1s
e if 0 t 2
Q(t, s) =
g( 2t1+s
1+s
) if 1s
2
t1

Corollary 2.1 The fundamental group of a topological group is abelian and there is
an isomorphism between (G, e)/ and (G, e)/(G, e)

Proof. To prove it is abelian we use the same idea as in the previous proof that is:

[f ] [g] = [e f ] [g e] = [e g] [f e] = [g] [f ].

39
2.4 The fundamental group of SO(3)
Widely used in physics and actually very popular in computer graphics are the rotation
groups. We have already calculated the fundamental group of S 1 , (S 1 )
= Z, which
is isomorphic to SO(2). Now it is very natural to move to higher dimension. The
space SO(3) has a fascinating topological property, there exist loops that cannot be
continuously deformed to the trivial path, but going twice along such a loop gives
another loop, which is deformable to the trivial one. For example, if you rotate an
object in R3 by 2 along some axis, you get a motion that is not deformable to the
trivial motion (no motion at all), but a rotation by 4 is deformable to the trivial
motion. Further, a rotation by 2 along any axis can be deformed to a rotation by 2
along any other axis. We shall call a full rotation in R3 any motion that corresponds
to a loop in SO(3), based on the identity. Thus, it turns out that there are two
classes of full rotations: topologically trivial (deformable to the trivial motion) and
topologically nontrivial. Every nontrivial full rotation can be deformed to any other
nontrivial full rotation. Two consecutive nontrivial full rotations produce a trivial one
[2]. This results to the fact that (SO(3)
= Z2 . The idea of the proof is to show, the
fundamental group of SO(3) has 2 elements. This will be done by using the Theorem
2.11 and 2.12.

Theorem 2.11 Let p : C B be a covering map such that p(c0 ) = b0 . Let f and
g be two paths in B beginning at b0 and ending at b1 . Let fe and ge be their respective
liftings to paths in C beginning at c0 . If f and g are path homotopic, then fe and ge are
path homotopic and end at the same point of C. [7]

Proof. Let f and g be homotopic and let F : I I B be the homotopy between


them. Since the paths begin at b0 we know that F (0, 0) = b0 . Let Fe : I I C be
the lifting of F to C. Then we know that Fe(0, 0) = c0 . By Lemma 8, Fe is a path
homotopy, such that Fe(0 I) = {c0 } and Fe(1 I) is a one-point set {c1 }.
Consider the restriction map Fe|Y {0} of Fe to the bottom edge of I I. This is a path
in C beginning at c0 that is a lifting of F |Y {0} . Since lifting of paths are unique,
we must have Fe(s, 0) = fe(s). Similarly, Fe|Y {1} is a path on C that is a lifting of
F |Y {1} , and it begins at c0 because Fe(0 I) = {c0 } . Again, since liftings are unique,
Fe(s, 1) = ge(s). Hence, both fe and ge end at c1 , and Fe is a homotopy between them.

Definition 2.3 Let p : C B be a covering map such that p(c0 ) = b0 . Given an


element [f ] of (B, b0 ), let fe be the lifting of f to a path in C that begins at c0 . Let
the map : (B, b0 ) p1 (b0 ) be such that ([f ]) denotes the end point fe(1) of fe.

40
Then is a well-defined set map. We call the lifting correspondence derived from
the covering map p.[7]

Theorem 2.12 Let p : C B be covering map such that p(c0 ) = b0 . If C is path


connected, then the lifting correspondence : (B, b0 ) p1 (b0 ) is surjective. If C is
simply connected (every loop is contractible to a point) , it is bijective.[7]

Proof. Let C be path connected, and let c1 p1 (b0 ). Then there is a path fe in C
from c0 to c1 . Then f = p fe is a loop in B at b0 and ([f ]) = c1 by Definition 2.3.
Suppose C is simply connected. Let [f ] and [g] be two elements of (B, b0 ) such that
([f ]) = ([g]). Let fe and ge be the liftings of f and g, respectively, to paths in C that
begin at c0 . Then, by Theorem 2.11, fe(1) = ge(1). Then, since C is simply connected,
there is a path homotopy Fe in C between fe and ge. Then p Fe is a path homotopy in
B between f and g. Since simply connectedness implies path connectedness we have
that is both surjective and injective; hence we have a bijective correspondence.

Now look at the commutative diagram on Figure 17. For us, using Theorem 2.12,
B = SO(3) and C = S 3 . We do not know p explicitly, but we will construct it using

Figure 17: Commutative diagram representing that S 3 is a double cover of SO(3) via
the isomorphism between RP 3 and SO(3). [6]

and pe so that p = pe . We will show that S 3 is a double cover of RP 3 and that there
is an isomorphism between RP 3 = S 3 \{1, 1} and SO(3), so that the Theorem 2.12
is applicable. Since S 3 is simply connected, is bijective and we will see that p1 (b0 )
is bijective and that it is a two element set.
Let us talk about what is RP 3 first.

Definition 2.4 The projective space RP 3 is the quotient space obtained from S 3 by
identifying each point q S 3 with its antipodal point q.

41
Then is a natural projection of quotient space. That is (q) = q when q is in the
upper half of the sphere without half of the equator, (q) = q otherwise. From the
construction it is obvious that S 3 is a double cover of RP 3 . Now let us move to the
part that RP 3 is isomorphic to SO(3). We will do that with the help of quaternions,
so we will continue by briefly reminding the reader of quaternions using [8].

Definition 2.5 Quaternions form a four-dimensional associative normed division al-


gebra over the real numbers, denoted H, and represents a four dimensional real vector
space R4 with basis 1, i, j and k.
The basis satisfies the identities i2 = j 2 = k 2 = ijk = 1.

Definition 2.6 A quaternion q H is written as q = q0 + q1 i + q2 j + q3 k and his


conjugate is defined as q = q0 q1 i q2 j q3 k for q0 , q1 , q2 , q3 R.

Definition 2.7 An element of the form q0 +0i+0j +0k is called a real quaternion, and
an element of the form 0+q1 i+q2 j+q3 k, is called a pure quaternion. If q0 +q1 i+q2 j+q3 k
is any quaternion, then q0 is called the scalar part, and q1 i + q2 j + q3 k is called the
vector part.
Then we can write any quaternion as q0 + ~q.

Then addition and multiplication are as follows:


p + q = (p0 , p~) + (q0 , ~q) = (p0 + q0 , p~ + ~q),
pq = (p0 , p~)(q0 , ~q) = (p0 q0 p~ ~q, p0 ~q + q0 p~ + p~ ~q). where is the dot product, and
is the cross product.

Definition 2.8 The reciprocal of q is defined to be


q
q 1 = (2.5)
kqk2
where kqk is the standard Euclidean norm.

Note that for unit quaternions (kqk = 1) q 1 = q . We can identify the set of all unit
quaternions with S 3 , as we have already used the notation q previously. In particular,
S 3 forms a group (a Lie group) under multiplication: S 3 = {q H |qq = 1}, where
H = H\{0} is the multiplicative group of all non-zero quaternions. By representing
a unit quaternion on the form q = (q0 , ~q) , we can think of the vector part as being
the axis about which a rotation occurs, and the scalar part as the amount of rotation
that occurs about the given axis. In fact, there is an explicit expression of a unit
quaternions, representing a rotation of an angle around a unit vector u R3 , given
by q = cos 2 + u sin 2 .

42
Lemma 9. Let R, u S 2 to be an imaginary unit and q = cos 2 + u sin 2 to be a
nonzero quaternion. Then the map p : R3 R3 , defined as pq = qrq 1 represents a
rotation of an angle in R3

Theorem 2.13 SO(3)


= S 3 \{1, 1}.

The proofs of Theorem 2.13 and Lemma 9 can be found in [8]. With this informa-
tion we have completed the commutative diagram on Figure 17.

Theorem 2.14 The fundamental group of SO(3) is Z2 . [6]

Proof. We represent the rotation (SO(3)) with unit quaternions. A rotation through
angles between 0 and corresponds to the point on the opposite side of the origin, at
the same distance from the origin. This can be seen from the fact that the rotation is
expressed as pq (r) = qrq 1 and we have pq (r) = (q)r(q)1 = qrq 1 which implies
that q is mapped to the same rotation. Then, either we rotate through an angle of
, or we rotate through an angle of , we still get the same rotation. Then we have
a two-to-one map and a continuous function

p : S 3 SO(3)

given by p(q) = pq .
Moreover, p is a group homomorphism:

pq1 q2 (r) = q1 q2 rq11 q21

pq1 pq2 (r) = pq1 (pq2 )(r) = pq1 (q2 rq21 ) = q1 q2 rq21 q11

Hence pq1 q2 = pq1 pq2 and the structure is preserved.


By the map pq (r) = qrq 1 we easily observe that p(q) = I q = 1. Then by the
first isomorphism theorem, the induced map

pe : S 3 /{1, 1} SO(3)

is a group isomorphism.
We know that the projective map : S 3 RP 3
= S 3 /{1, 1} is a double covering
map identifying antipodal points of S 3 . Hence, by the commutative diagram on Figure
17, p = pe is a double covering map, identifying antipodal points on the surface of S 3 .
We now use Theorem 2.12. Since S 3 is simply connected, the lifting correspondence

: (SO(3)) p1 (pq )

43
is a bijective correspondence. Since p1 (pq ) is a two element set, we know that
(SO(3)) is a group of order two.
Let G = {e, a} be a group of order two, and let e be the identity element. Then we have
the following: Since e is the identity element, we have e x = x e = x for all x e, a.
Since a is a nontrivial element, a has an inverse a1 such that a1 a = a a1 = e.
In our case, a1 must be either e or a. Since a1 = e does not work, we must have
a1 = a. We know that Z2 = {0, 1} under addition modulo 2 is a group. By our
arguments, e must be replaced by 0, and a by 1.
This shows that the fundamental group of SO(3) is Z2 .

44
Conclusion
The specific topological property, that fundamental group of SO(3) is Z2 , plays a
fundamental role in our physical world. To the two homotopy classes of closed paths
in SO(3) correspond precisely two principally different types of elementary particles:
bosons, with integer spin, and fermions, with half-integer spin, having very distinct
physical properties. The difference can be traced to the fact that the complex (possi-
bly multicomponent) wave function determining the quantum state of a boson is left
unchanged by a rotation by 2 of the coordinate system while the same transformation
multiplies the wave function of a fermion by 1. This is possible since only the mod-
ulus of the wave function has a direct physical meaning, so measurable quantities are
left invariant under a full rotation by 2. However, as discovered by Pauli and Dirac,
one needs to use wave functions having this (unexpected) transformation property for
the correct description of particles with half-integer spin, such as the electron [2]. We
will demonstrate this by Diracs scissors experiment. P.A.M. Dirac perfmormed this
experiment at his lectures in the 1930s.

DIRACS SCISSORS EXPERIMENT [6] We want to consider an element of


the group. A generator for (SO(3)) is any closed curve obtained by projecting any
curve connecting antipodal points in S 3 down to SO(3). Consider the semicircle
{cos + i sin S 3 |0 }, the intersection of S 3 with the upper half of the
xy-plane. This maps by p to a loop in SO(3) and represents a nontrivial element of
(SO(3)). We look at what it does to an arbitrary element in R. Let r = xi + yj + zk
and q = cos + i sin . Then we have pq (r) = qrq 1 = ix + j(y cos 2 z sin 2) +
k(z cos 2 + y sin 2). Hence, it fixes the complex plane on so only acts on the jk-plane.
For q = yj + zk we have pq (r) = qrq 1 = j(x cos 2 y sin 2) + k(y cos 2 + x sin 2)
which tells us that q = cos + i sin acts on the jk-plane as to rotate through 2.
Since [0, ], it follows that the nontrivial element of (SO(3)) is given by a path
of rotations about a given axis whose angle [0, 2]. Further we let the rotation to be
around the z-axis and let r(t) be counterclockwise rotation in the xy-plane through the
angle 2t, r a loop representing the nontrivial element of (SO(3)). Since (SO(3))
is a group of order two, it follows that the loop f = r r must be homotopic to the
constant loop e SO(3), f a loop whose value f (t) is the rotation of angle 4t. Let
f be this loop, and let [f ] (SO(3)).

45
Since [f ] = [e] we have a homotopy F : I I SO(3) such that

1. F (s, 0) = e and F (s, 1) = f (s),

2. F (0, t) = e and F (1, t) = e.

With this set up we are ready to demonstrate the experiment.


Take two threads and attach one end of them to the two inside rings of a pair of
scissors. Attach the other ends to some fixed objects in the room, e.g. two chairs.
Then rotate the scissors one complete turn. Then, while keeping the object stationary,
try to untangle the threads. Although patience is a good thing, it will not help you,
you will either way be unable to do this no matter how long you try. However, if
you rotate the scissors one more time in the same direction, that means two complete
turns in total, you will actually be able to untangle them, keeping in mind to hold
the scissors stationary as you are doing it. Then what does this tell us? Well, by
rotation not once, but two complete turns around the same axis, we got back to where
we started, and hence we have a cyclic group of order two, whose generator is a loop
of rotation around an axis of angle 2.

Figure 18: Diracs scissors experiment [6]

46
References
[1] GOWERS John. (http://math.stackexchange.com/users/26267/john-gowers),
What is the difference between homotopy and homeomorphism?. URL (version:
2015-01-21), http://math.stackexchange.com/q/491039

[2] HAJDINI Ina - STOYTCHEV Orlin. The Fundamental Group of SO(n) Via Quo-
tients of Braid Groups. arXiv:1607.05876v1, 2016.

[3] HATCHER Allen. Algebraic topology. Cambridge: Cambridge University Press,


2002. ISBN 978-0521795401

[4] KNAPP W. Antony. Advanced Real Analysis. Springer Science & Business Media:
New York, 2008. ISBN 978-0817643829

[5] KOSNIOWSKI Czes. A first course in algebraic topology. Cambridge: Cambridge


University Press, 1980. ISBN 978-0521298643

[6] MORK Eirik. The Fundamental Group of SO(3). Norwegian University of Science
and Technology, 2014. http://hdl.handle.net/11250/2352611

[7] MUNKRES James. Topology (2nd Edition). Upper Saddle River: Prentince Hall,
2000. ISBN 978-0131816299

[8] ZLATOS Pavol. Linearna algebra a geometria. Bratislava: Marencin PT, 2011.
ISBN 978-80-8114-111-9

47

You might also like