INDEX

S.No. Topic Page No.

WEEK- 1

1 Lecture – 01 1
2 Lectwure – 02 8
3 Lecture – 03 15
4 Lecture -04 18
5 Lecture -05 20
6 22
Lecture -06

WEEK- 2

7 26
Lecture -07
8 Lecture -08 37
9 Lecture -09 45
10 Lecture -10 48
11 Lecture -11 49
12 Lecture -12 54

WEEK- 3

13 Lecture -13 59
14 Lecture -14 64
15 Lecture -15 69
16 Lecture -16 74
17 Lecture -17 79
18 Lecture -18 82

WEEK- 4

19 Lecture -19 87
20 Lecture -20 92
21 Lecture -21 99
22 Lecture -22 103
23 Lecture -23 110
24 Lecture -24 118
 WEEK-5

25 Lecture -25 124

26 Lecture -26 130
27 Lecture -27 135
28 Lecture -28 141
29 Lecture -29 148
30 Lecture -30 153

WEEK- 6

31 Lecture -31 161

32 Lecture -32 166
33 Lecture -33 173
34 Lecture -34 178
35 Lecture -35 182
36 Lecture -36 191

WEEK- 7

37 Lecture -37 196

38 Lecture -38 202
39 Lecture -39 207
40 Lecture -40 220
41 Lecture -41 233
42 Lecture -42 235

WEEK- 8

43 Lecture -43 239

44 Lecture -44 248
45 Lecture -45 251
 NPTEL

NPTEL ONLINE COURSE

Introduction to Abstract

Group Theory

MODULE – 01

Lecture – 01 - “Motivational Examples of Groups”

PROF.KRISHNA HANUMANTHU

CHENNAI MATHEMATICAL INSTITUTE

I’m Krishna Hanumanthu from Chennai Mathematical Institute. I’m going to teach a course on
abstract group theory over the next 8 weeks. My goal here is to introduce the basic ideas, give
some motivation, and do some standard theorems in the subject, and to explain a little bit about
why groups are very important in various areas of mathematics and even outside. So today I
want to start off by first giving you some examples which I hope will motivate the definition,
then I’ll give you the definition and look at some properties.

Okay, so the course is on introduction to abstract group theory, okay. Groups are algebraic
objects which abstract, as the word suggest, certain important features of well-known
mathematical objects. So I want to do some examples which illustrate this, and in each example I
want to point out the crucial piece of information that we want to retain, okay.

So the first and most important example of a group that we all know is the group of integers, so
this Z is the symbol for this, so this is the set, let us start with this, it is the set of all integers, so
integers are, for example these are all negative integers, 0, and positive integers so it’s an infinite
set, consisting of these numbers, okay.

Groups are sets along with a certain operation on them, so and the most familiar operation for us
on Z is the addition, so what is this? Addition, we all know how to add two integers, and what
are the properties of addition, let’s try to identify this that we want to abstract out and define a
group later, so what is the property of addition? So given any two integers we can add them and
we get another integer, so if you add, if we add two integers we get another integer, so this is

the starting point, so we say that adding integer is the binary operation on the set of integers, so
I’ll define this in more formal setting later, but it simply means the word binary refers to the fact

1
that we add two integers to get a third integer, so for example if you add 3 and 2 you get 5, if you
add 5 and -2 we get 3, so if you perform the operation of addition on two integers we get another,

the output is 1 integer, so input is 2 integers, operation is addition output is 1 integer, so that is
the first property of addition.

So let’s look at another important property of addition, there is a special element called 0, 0 is a
special element, so what do I mean by that? So let me say 0 as a special element inside which is
to say special in the following sense, if I add if N is any integers, so let us take an arbitrary
element N, then if I do N+0 I get N back, which is also same as 0+N, 0 is the only integer with
this property, if you add 0 to this N integer you get that integer back, no other integer has this
property if you add 1 to an integer you do not get the integer back, so 0 is called the identity

element, identity element for addition, okay, so this is the second property. So remember again
the first property was if you add two integers you get another integer, you have a special element
that we call identity element which has the property that if you add this to any element you get
the element back, you do nothing in another words.

The third important property that I want to identify is every element which in this case is an

integer, has an inverse. What is an inverse? Inverse is an element that you can add to get the
identity element, so for example if you, what element do we want to add to 5? So we want to add
something to 5 to get 0, what is that element? That element is obviously is -5, if you want to add
something to -3 to get 0, you will add 3, more generally if you want to add –N to N you get 0, so
this is an inverse, so inverse is the opposite so in some sense, if you add inverse to an element
you get the identity element which I have already declared a special element called the identity

element. So every element has an inverse, so we have identified 3 properties of addition on

integers, if you add two integers you get another integer, there is a special element called 0, and
every element has an inverse.

Finally, I want to note the following property of addition, so remember addition is a binary
operation, so in other words we can add two elements, if you add 3 and 2 you get 5, but we
cannot add a priori 3 elements, how do you add 3 elements? What is the meaning of this?

Remember addition is only a binary operation, given 2 things we produce 1 output, but if you are
given 3 things there are 2 possible ways of doing the addition, so if I asked you what is 3+2+5
you would first, one option is to combine 3 and 2 first, and then this now 3+2 is again an integer,
and add 5 to it, so this is 3+2 in bracket, that means you apply the binary operation to 3 and 2
and for which you get 5, and then you add 5 so that gives you 10. But you can also do the
following, you can first add 2 and 5, the point is to group two things together, in the first option I
grouped 3 and 2, in the second option I’ll group 2 and 5, so then I get 3+7 which is also 10, so I
get the same answer.

2
So no matter how I group elements, I get the same answer which is very important, otherwise
there will be an ambiguity on how to add 3 elements or 4 elements or 5 elements, so this is called
associativity of addition.

Associativity means we can group things again I’ll do this more formally later, but we can group,
given 3 elements we can group 2 of them together in two different ways, and we get the same
answer, okay, so this is the fourth property of addition on integers, so we can unambiguously add
3 elements so we can use the associativity property of addition.

So just to recall properties of addition on integers: one it is closed, so I’m going to use this word
from now on, it is closed under addition which is, whenever we say a property is closed that
means if I apply that property I’m within the same set so, if I take 2 integers in other words 2
elements of Z perform addition I land again in Z, there is an identity, the second property is there
is an identity element, namely 0, this 0 is an identity element, every element has an inverse
namely –N is the inverse of N, addition is associativite is the last property, so these are the four
properties of addition on Z.

If you think about it a little bit there is another property of addition that I did not highlight now,
but which we will come to later, so I’ll record it here and we will come back to it later, another
property of addition is, for example if I add 3+6, I get 9, which is same as 6+3, so the order in
which I perform the binary operation, first 3 then 6 or first 6 then 3 I get the same answer, this is
not such an important property as the first 4, but let us record this for the moment and then we
will come back to it later, so this are the properties of integers, addition on integers that I want to
record.

And I want you to remember that these properties while they are so obvious for addition they’re
not necessarily always true, for example if you take multiplication on integers, there is no inverse
for multiplication if you think about it, because if you multiply, what is the inverse under
operation of multiplication on integers, for example 2 has inverse 1 by 2 which is not an integer,
so it’s not clear that these properties always hold.

So with that let’s go to the second example, which again remember I’m trying to give three
examples to motivate the definition of a group, so the second example that I want to give is the
following, so this is not so obvious, but I would like you to pay close attention to this, so let’s
take the set consisting of 1, 2, 3, okay, so these are 3 elements, so I’m not going to use any

3
property of 1, 2, 3 other than that they are 3 different elements of a set, so you all know what is
the bijection of this set so I’m going to consider bijections of 1, 2, 3. A quick recap of the
definition of a bijection, what is the bijection? A bijection is a function from {1, 2, 3} to {1, 2, 3}
which is 1-1 and onto, okay. 1-1 and onto are properties of functions of sets, 1-1 means two
different elements of the domain, so this is the domain, this is the codomain, two different
elements of the domain go to two different elements of codomain that’s 1-1, onto means every
element of the codomain is equal to, an image of something from the domain so codomain is
equal to the range of this function. So for example, so I’m going to use this notation to denote
functions, function 1 going to 1, 2 going to 2, 3 going to 3 is a bijection, clearly, because two
different elements everything in the set 1, 2, 3 is in the image, so it’s a bijection, it’s in fact the
identity map.

On the other hand if you send 1 to 1, 2 to 1, and 3 to 2 this is not bijection, because 1 and 2 are
different elements that both go to 1, so it’s not 1-1, of course also it’s not onto, because 3 is not
in the image, so I’m not interested in functions which are not bijections, so as I said bijection is a
1-1 and onto map.

Now I’m going a define a different set, so define, which I’ll denote by S sub 3, S sub 3, the 3
refers to the fact that I have 3 elements in my starting set, so S3 is the set of all bijections from
{1, 2, 3} to {1, 2, 3}. So this is the set of all 1-1 and onto functions from this set to that set, so
this is my set now, so I started with the set 1, 2, 3 consisting of 3 elements 1, 2, 3 and I’ve
defined a new set now consisting of all bijections from 1, 2, 3 to itself, so I would like to define
an operation on S sub 3. But before that let me first list all the elements of S3 it’s not very big,
for example the function that I defined earlier as an example of a bijection 1 going to 1, 2 going
to 2, 3 going to 3 is a bijection let’s call that F1.

So what are the elements of S3? So I’m going to try to list all the elements of S3, so the first
function because these are elements are functions, I’m going to use F with a subscript to denote
this elements, F1 is 1 going to 1, 2 going to 2, 3 going to 3 so that’s F1, this is in fact the identity
function, meaning 1 goes to 1, 2 goes to 2, 3 goes to 3.

What are the other bijections? You can also send let’s call it F2, you can send 1 to 2, and 2 to 1,
3 to 3. Note that this is also a bijection, because different elements go to different elements,
everything in the set is a image, 2, 1, 3 are all in the image. And if you think about it what I have
done is I’ve just interchanged 1 and 2 while keeping 3 fixed, 1 goes to 2, 2 to goes 1, but 3 goes
to 3, similarly I can interchange, I can interchange 1 and 3 so that means I send 1 to 3, 3 to 1 and
2 to 2, nothing happens to 2, but 1 goes to 3, 3 goes to 1 so this I call F3.

4
What are some other bijections? So F3 is done, so now F4 let’s say I interchange 2 and 3, so 1
goes to 2, 2 goes to 3, 3 goes to 2, so here I do not move 1, but 2 goes to 3, 3 goes to 2 that’s F4.

Are there any other bijections? There are, for example if you take if one of the elements is fixed
which is to say in this, in F1, F2, 1, 2, and 3 are all fixed, in F2 3 is fixed, F3 2 is fixed, and in F4
1 is fixed, if you think about it if an element is fixed, if a function fixes either 1, 2, or 3, it must
be F1, F2, F3 or F4, so the remaining bijections if there are any, must not fixed anything so, and
there are such things, for example 1 can go to 2, 2 can go to 3, 3 can go to 1, neither 1 nor 2, nor
3 is fixed so everything changes, so 1 goes to 2, 2 goes to 3, 3 goes to 1, and this is a bijection.

Similarly F6 also does not fix anything, it sends 1 to 3, 2 to 1, and 3 to 2 that’s also a bijection,
and some thinking on your part tells you that this are all the bijections, there are only six of them,
so S3 is the set consisting of F1, F2, F3, F4, F5, and F6, so I’m going to use these fixed specific
definitions later, so please note that F1, F2, F3, F4, F5, F6 are the ones I define just now, so F1 is
the identity element, F2, F3, F4 fix exactly one element of the set 1, 2, 3, F5 and F6 do not fix
anything.

Now remember that when we look at the previous example of integers, the set was easier there,
just the set of integers, but there we looked at addition, here what do you want to look at? So we
want to get hold of an operation on this set, so in other words given two functions we want to
produce another function, there is an obvious candidate to do this, we have all learned in school,
if you take 2 functions we can compose them, in this case because the domain and codomain are
the same we can always compose any two functions, so consider the operation composition of
functions on S3. If you think about it, if you take 2 bijective functions and you compose them
you get another bijective function, so it must be again in S3.

So let us just walk this out more concretely in a specific example, so for example if you do F2
circle F5, so remember composition is usually denoted by circle and in practice I’ll just omit the
circle and write this as F2, when there is nothing I’ll read it as F2 composed with F5, so what
does this do? Remember I apply F5 first then apply F2, F5 if you now see here, F5 sends 1 to 2,

and F2 if you look at the previous page F2 sends 2 to 1, so F5, F2 composed with F5 sends 1 to
1, because F5 sends 1 to 2, and F2 sends 2 to 1. Similarly you can just look at this 2 goes to 3
under this, and 3 goes to 2 under this.

And remember if you look back at the list of elements that we have listed in S3, this is nothing
but this is same as F4, so this is an example of a, what I mean by the operation being closed, so if
I take 2 functions, F2 and F5 in S3 I compose them I get another function of, another element of
S3.

Now let’s do F5 composed with F2, now remember I first apply F2 and then I apply F5, and if
you do this as before you can quickly check that this sends 1 to 3, 2 to 2, and 3 to 1, and this is

5
same as F3, if you recall when I did the example of integers I did, I noted that after listing the 4
important properties I said that integers also have a fifth property, namely the order in which I
apply addition does not matter, when I add 3 to 5 I get 8, but when I add 5 to 3 I also get 8, and
as this example suggests as the new example suggest order does matter here, if I do F2 composed
with F5 I get F4, whereas if I do F5 composed with F2 I get F3, so order in which I compose
gives me different results, so however I can say that composition is a binary operation, so I’ll say
composition, so composition is a binary operation which is same as saying, this is same as saying
that S3 is closed, see note that I’ve not proved this, I’ve only checked that the composition is
closed at least for 2 elements, I have 2 composed with F5 is again a S3, F5 composed with F2 is
also in

S3, but here I am saying that if you take any two elements compose them I get another element
of S3, and this is an exercise that you can do and maybe later I will tell you how to do this, it’s
not clear but this is an easy exercise that composition of two bijections is also a bijection. This is
the exercise that will make justify the statement that composition is a binary operation on S3 or
that S3 is closed under composition, because if I, what is S3? Again let’s recall, S3 is the set of
all

bijections from the set {1, 2, 3} to itself, so if you composed two elements of it, in another words
if you compose two bijections you get another bijection from 1, 2, 3 to 1, 2, 3, so S3 is closed
under composition, so S3 with composition has the property that integers with addition have, so
this is the first property remember.

What is the second property? There is an identity, again recall, in the first example 0 is the
identity element of Z under addition, here what is the identity element of S3, it is simply F1,
remember F1 is the identity function from 1, 2, 3 to 1, 2, 3, so for example F2 composed with F1
is F2. Similarly F3 composed with F1 is F3 and so on. This is clear because F1 is the identity
element, so composing with F1 does not change the function at all, so there is an identity
element.

Now the third property, interesting property, is there an inverse? Is there an inverse for every
element of S3? There is, for example if you do F1 composed with F1 which is often for
convenience of notation denoted as F1 squared, what is F1 composed with F1? If you now go
back and see, if you go back and see F1 sends 1 to 1, 2 to 2, sorry actually I don’t want to do F1

squared, F1 squared is simply F1 that is obvious, I want to do F2 squared, what does F2 squared
do? 1 goes to 2, but 2 goes to 1, so F2 squared sends 1 to 1, 2 goes to 1 under F2 but by repeating
F2 we get 1 goes to 2, so 2 goes to 2 under F2 squared, and 3 anywhere does not move under F2
so 3 goes to 3 under F2 squared also, so what I want to do is, F2 squared is F1, similarly if you

6
quickly check F3 squared as well as F4 squared is F1 which remember is the identity element
from the previous page, remember that F1 is the identity element, and in an inverse is an element

such that when you multiply, I use the word multiply here but I should really use composed, if I
compose F2 with F2 I get identity, if I compose F3 with F3 I get identity, if I compose F4 with
F4 I get identity.

What is an inverse for F5? If you go back to the definition of F5 you will see quickly a simple
calculation tells you that F5 composed with F6 is the identity element, because you have the
definition of F5 and F6 here, 1 goes to 3 under F6, 3 goes to 1 under F5, so under F5 composed
with F6 1 goes to 1, similarly 2 goes to 2, and 3 goes to 3, so F5 composed with F6 is F1, so in
other words, in words F6 is the inverse of F5, and also F6, F5 is the inverse of F6, so every
element has an inverse.

And the fourth property is the composition of function associative? which it is, composition of
functions is associative, this is something that you perhaps have done in the past, it’s an easy
check that if you compose two functions and then compose the third function is same as this,

okay, so composition of functions is associative, so that comes easily here, so again the upshot
here is S3 which was defined to be the set of bijections from {1, 2, 3} to {1, 2, 3} namely F1, F2,
F3,

F4, F5, F6 under the composition is closed under composition there is an identity element, every
element has an inverse and composition is associative, so and I want to point out that the
operation here is composition, it’s not addition like in the case of integers. So one more quick
example I want to do, third example before we formally define groups.

Third example which is also one of the motivating examples for groups is the following, so
consider an equilateral triangle, so what is an equilateral triangle, so this is a triangle with all
three sides congruent, so roughly it looks like this, so I want to consider rotational symmetries of
this, so what do I mean by rotational symmetries? Rotational symmetries are the following, so I

want to rotate the plane on which this triangle lives, and the rotational symmetry 1 is one where
after, if I perform a rotation I get the triangle back, so rotations happen around a point, so let’s
say I rotate around the point in, median point of this triangle, so and rotation is always
determined by an angle, so if at home you can practice this by cutting out an equilateral triangle
from a paper and just rotating it, so if you rotate, let’s call this side vertices A, B, C, if you rotate
what rotations preserve the triangle, so this labeling is just for convenience, that’s not part of the
data for equilateral triangle, if I rotate what happens? So for example if I rotate by, so let say I’m
rotating anti-clockwise, if I rotate by 90 degrees you can quickly see I do not get the, so this is
the starting point, this is the starting triangle, if I rotate by 90 degrees I get something like this,

7
roughly something like this, so C will move here, A will move here, B will move here, this is not
same as, clearly this is not same as, not same, 90 degree rotation, okay. So this is actually, yeah,
this is not a symmetry.

So I don’t want to spend lot of time on what are the rotations which preserve the triangle, but the
following happens, so if you rotate by 120 degrees, or rotate by 240 degrees, of course you can
also rotate by 360 degrees, which means we are not doing anything, these are the rotational
symmetries of an equilateral triangle, so let us denote these by some names, so for example I call
this R1, I call this R2, I call this R3, and let’s take the group or the set G to be {R1, R2, R3}. I’m
doing this example so that you get an idea of different kinds of groups that we encounter in
mathematics, so this is the set, and this set is simply the set of all rotational symmetries of an
equilateral triangle.

And as the previous two examples already indicated, we need an operation on this set, what is
the operation? Operation is again composition, so operation is again composition so I can
compose two rotations and if you think about it, if you compose two rotations you get another
rotation, in this case it’s very simple if you rotate first by R2, and then by R3 if you first rotate
by 120 degrees and then by 240 degrees, remember I usually do not write the composition
symbol, first rotate by 120 degrees then by 240 degrees, cumulatively I’m doing a rotation by
360 degrees, so I get R1.

Similarly if I do compose by 240 degrees, sorry 120 degrees then again by 120 degrees then I’m
composing by 240 degrees, so I get R3. Similarly R3 composed with R3, if I compose by, rotate
by 240 degrees then by 240 degrees so I’m rotating by 480 degrees which is rotation by 120
degrees, so it’s R2, so as you can list all the possible binary operations, so composition is a

binary operation on G, so the first property is true. Second property is that there is an identity,
namely R1, remember R1 is the rotation by 360 degrees or 0 degrees, performing it does not do
anything so R2 composed with R1 is R2, and R3 composed with R1 is R3. Every element G has
an inverse, has an inverse, as the previous page shows R3 composed with R2 is R1, so R3 is the
inverse of R2, and R2 is the inverse of R1, and as again as before composition is associative, just
like the second example composition is associative, composition of functions is always
associative so that comes here for free, so again just like the examples 1 and 2, the third example
also we have our set G on which composition is a binary operation, there is an identity element,
every element has an inverse and composition is associative, okay, so these three examples I
hope give you some motivation to study groups and I’ll stop my first video now, in the second
video we will formally define groups and look at more examples.

8
 NPTEL

NPTEL ONLINE COURSE

Introduction to Abstract

Group Theory

MODULE – 01

Lectwure – 02- “Definition of a group and examples”

PROF.KRISHNA HANUMANTHU

CHENNAI MATHEMATICAL INSTITUTE

So in the previous video we saw some examples of sets with operations, we haven’t yet define
what a group is, so these were sets with operations, for example the first one was integers under
addition, the second one was the set was the bijections from {1, 2, 3} to {1, 2, 3} under
composition, and the third example was rotational symmetries of an equilateral triangle, so in all
these examples we saw that the binary operation at certain properties.

So hopefully those examples motivated the definition of group, go ahead and formally define a
groups, so this is the definition of a group, which is the object that we will study in this course,
okay, so this should now be clear after the examples that we have done, so a group is a set, let’s
say G, we denote groups by usually G, with a binary operation, so what is a binary operation? So
it is a function from G x G to G, so it is just a mathematical way of writing, what we discussed in
the previous video, so given two elements of the group, so G x G is the set Cartesian product of
two sets, so elements are pairs of elements of G, there is a way to produce another, just for clarity
let me use star as my operation, though in practice we will not use this, so star is the binary
operation.

So remember that binary operation definition already includes the statement that the operation is
closed, because you take two elements of the group G, perform the star operation you get a third
element of the group, g1 star g2, satisfying the following properties, remember that the examples
that we studied in the previous video, the operation had four properties, namely that it is closed
or that it is a binary operation, that’s already listed here, so the remaining three properties are
there is an identity element which is denoted usually by e, what is the property of the identity
element? g star e is equal to e star g and you must get g back, so remember this is exactly the
identity element that we studied, 0 was the identity element for the operation addition on
integers, but we want to insist that whether you compose g with e or e with g you get g back, so
that is the first property.

9
Every element g of the group set G has an inverse, which is, if g is an element there is something
called g inverse, we denote this by g inverse, so the inverse is denoted by g inverse that’s just
convenience, what is the property of g inverse? g star g inverse is same as g inverse star g and
you should get the identity element back, okay, so remember that first identity element must be
there then inverse can be defined.

And what is the third property of the binary operation? Star is associative, which is to say that g1
star g2, star g3 must equal g1 star, g2 star g3, this is true for all elements g1, g2, g3 in G. This is
true for every triple of elements of the group, so then we say that the operation is associative, so
remember we always have to emphasize that a group is a set G along with a

binary operation, a set itself is not a group, binary operation is important to define a group, so we
say the correct way to say this is G, star is a group, that’s all, so a group is a set along with a
binary operation star which admits an identity every element of the set has an inverse and the
operation is associative, so G star is a group, so again let me emphasize star is important.

So examples, let’s look at examples, remember that in the previous video we looked at some
examples, and my goal has been always to abstract the key properties in those examples, and the
key properties in this examples was the operation was closed namely it is a binary operation,
there is an identity element and every element has an inverse and it is associative, so I have
abstracted out those properties and defined a group, this is what we now call an abstract group,
so “abstract” refers to the fact that it is not specific, it’s not integers that you are used to, it’s not
functions that you’re used to, it’s not real numbers that you are used to, the set along with the
operation has no structure other than that imposed by this operation and these properties, that’s
why it’s an abstract group, so integers with addition is a group, this is our first example in the
previous video.

Similarly S3 along with composition is a group this is composition, so plus is the addition of
integers, S3 with composition of functions. Similarly I defined rotational symmetries of an
equilateral triangle that we call G earlier and again there the operation was composition, so these
three are examples of groups.

And as I said the operation is important to this, but whenever if there is no confusion, in other
words if the operation is clear from context .we write G is a group, instead of, okay, so it is
simpler sometimes to write that G is a group instead of specifying the operation because
sometimes we don’t need to specify the operation if you are looking at an example or a problem,
operation is clear from the context, okay, so I want to today discuss more examples. We have

10
already three examples, I want to discuss more examples so that we have a collection of groups
in mind before we study properties of these groups, let’s first look at some more examples.

Some obvious ones to begin with, so let’s take Q, these are standard notations that I’ll use
throughout the course, so this is the set of rational numbers, and under addition this is a group,
just like integers under addition is a group, rational numbers under addition is a group, and the
proof is similar, remember a group is a set with binary operation. In the case of Q the binary
operation was addition, so you can add two rational numbers, keep in mind that if you add two
rational numbers you get another rational number so it is a binary operation on the set of rational
numbers, there is an identity element namely the 0 element, so that’s an identity element, every
element has an inverse, what is an inverse of 1 by 2? It’s -1 by 2 so addition on rational numbers
has an inverse, and addition certainly is associative, so Q with addition is a group. Similarly R
which is the set of real numbers, as I mentioned these are going to be my standard notations, Z
refers to integers, Q with this bar here refers to rational numbers, R refers to real numbers, and
just like before R, + is a group, exactly for the same reasons, you can add two real numbers, you
get another real number. Similarly if you take the set of complex numbers, it’s also a group
under addition.

As you can see, unless you mention a possible binary operation the question of a group does not
arise, so you have to mention what the operation is, so under addition which is very familiar to
us, Z, Q, R, and C are all groups. Now let me ask you something, so let me study possibility of
groups under multiplication, so usual multiplication so, is for example, Z a group under, we have
seen this in detail in the previous video, it is a group under addition, but now I’m asking, is it a
group under multiplication, remember multiplication is a operation that we can perform on
integers. First property, is it closed? Here you have to ask yourself if I multiply two integers, do I
get another integer, yes, that’s okay, so of the required 4 properties the first property is okay,
multiplication is closed, because multiplying two integers, so if we multiply two integers, we get
another integer, so that’s okay. The second property is, is there an identity? Okay, so this is
where things get tricky, is there an identity element for multiplication? Think for a second, it
looks like there is, how about 1, the integer 1? So it looks like it is a multiplicative identity, so if
you do 3 times 1, this dot here for me is the symbol for multiplication, is 3. Similarly 5 times 1 is
5, 100 times 1 is 100, so is 1 multiplicative to identity? Actually it is not, it’s almost a
multiplicative identity but if you do 0 times 1, you get 0, so yeah, so actually it is a multiplicative
identity, that’s not a problem, so 1 is a multiplicative identity.

What about inverses? So here is where we have a problem, as I mentioned in the first video
briefly, what is an inverse of 2? What is the inverse of 2? Remember, inverse of 2 under

11
multiplication, so here I am interested in multiplication, additive inverse is -2 that was okay,
because under addition Z is a group and 2 has inverse, if you recall 2+ -2 was 0, but what is the
multiplicative inverse? You have to remember that definition of an inverse in a group, what is
the definition of an inverse? It is an element which has the property that if you perform the
binary operation you get the identity element, we agreed that, 1 is the multiplicative identity, so
which number can you multiply with 2 to get 1, so 2 times something should give 1, so what is
that? Certainly that has to be 1/2, but 1/2 is not an integer, okay, so in other words inverses do
not exist in Z for multiplication, they do exist for addition, so inverses do not exist in Z for
multiplication, so Z under multiplication ,this is multiplication, is not a group, okay, so Z under
addition is a group, but Z under multiplication is not a group, okay, so fine. Z is not a group,
because inverse is do not exist 1/2 is not an integer, okay.

Now let’s enlarge our set, so what about Q, under multiplication? Q under addition of course as
we saw earlier is a group, what about under multiplication? See here the problem of the integers
does not arise because 2 has an inverse now, namely 1/2, so 1/2 which was not an integer earlier
is nevertheless a rational number, so 2 has an inverse, remember inverse is must exist in the set,
again I go back to the definition of a group, every element g of G has an inverse, g inverse which
is again an element of the group G of the set G which has this property, for integers that is not
the case under multiplication. For rational numbers 2 has an inverse, but there is still a problem,
what is that problem? What is that? 2 has an inverse, but 0 does not have an inverse, why?
Because if you do 0 times anything and remember to be an inverse of 0 we must multiply with
something to get 1, but this is not possible, because if you multiply anything with 0 you get 0,
you can never get 1, so Q under addition, sorry Q under multiplication is not a group. Again Q
under addition is a group, but Q under multiplication is not a group because it seems very close
to being a group, because it is closed under multiplication certainly because if you multiply 2
rational numbers you do get

another rational number, there is an identity element namely 1, and almost all elements have
inverse, in fact every element other than 0 has an inverse, and certainly multiplication is
associative always, so but still it’s not a group because one element has no inverse, but if you
remove 0 from it, let’s denote Q star to be the set of nonzero rational numbers, in other words in
the notation of sets this is R, okay, so let me denote like this, this is Q – the element 0, if you
take nonzero rational numbers under multiplication they form a group, is it clear? Because
remember the only problem that we encountered in making Q under multiplication a group is
that 0 has no inverse, 0 does not have an inverse, but I’ve removed 0 here, 0 is not in Q star, so
we don’t ask for its inverse. Now if you take a rational number if r is in Q star, inverse of r,
remember which we denote by r inverse is simply 1/r, so it is now well defined because r is not
0, so Q star is a group, so I’ll let you go back to the definition and think a little bit if needed to
convince yourself that Q star has all the required properties, there is an identity element namely

12
1, every element has an inverse and multiplication is definitely associative, okay, so Q star under
addition, multiplication is a group.

And similarly R star which is as before set of nonzero real numbers, similarly C star set of
nonzero complex numbers are both groups under multiplication, because star means we have
removed 0, these are nonzero real numbers, nonzero complex numbers, and just like in the case
of rational numbers if I remove 0, multiplication becomes a group because every nonzero

element has an inverse under multiplication, identity element is still 1, multiplication is definitely
associative, so these are groups, but I want to remark that Z star which is the set of, if you define
it like this nonzero integers, Z star under multiplication is not a group, because here there was a
problem with even nonzero integers having inverses, not just the integer 0, because even 2 does
not have an inverse, so Z star is not a group, but R star, C star and Q star are groups under
addition, I also want to discuss more examples coming from this rational numbers, real numbers,
and complex numbers.

So quickly another example, if you define Q+ this is my notation, for positive rational numbers,
similarly R+ positive reals, are they groups under multiplication? If you think about it, are
groups under multiplication, because remember the multiplicative identity is 1 which is a
positive real number or rational number, so 1 belongs to Q+ and 1 belongs to R+, so I’m defining
R+ to be positive reals, Q+ to be positive rationals, we will keep using these in the course, they
contain 1, and inverse of a positive real number is again a positive real number, inverse of a
positive rational number is again a positive rational number, so and associativity comes for free
because multiplication of numbers is associative, and closed also comes for free because if you
multiply positive numbers you get positive number, so these are different examples of groups.

And I also remark once more that Q+ or R+ or Q star or R star or C star are not groups under
addition, they are groups under multiplication but not under addition because what is the identity
element for addition? As we have seen earlier it is the 0 element, but 0 is not in Q+ because it’s
not a positive rational number, it’s not in R+, Q star, R star, C star do not contain 0 because
obviously they are defined to be nonzero rationals, reals, and complex numbers, so they are not
groups under addition, but they are groups under multiplication. These examples make it very
clear that group operation is very important, obviously the name itself suggests that before you
want to understand a some set is a group or not, the sets have no meaning, unless you ask for
what is the operation, so under addition these are not even groups, under multiplication they’re
groups, okay, so it’s important to keep in mind that when we specify a group we specify, we
must specify a group operation, okay.

13
 NPTEL

NPTEL ONLINE COURSE

Introduction to Abstract

Group Theory

MODULE – 01

Lecture – 03- “More examples of groups”

PROF.KRISHNA HANUMANTHU

CHENNAI MATHEMATICAL INSTITUTE

So it’s important to keep in mind that when we specify a group, we specify, we must specify a
group operation, okay. So let’s look at couple of more examples of groups, before we study
properties of this, so I want to generalize the definition of S3 that we considered, so recall so
recall that S3 was the set of bijections of the set {1, 2, 3}, and S3 under composition is a group,
okay. We can generalize this, there is nothing special about 3. So now I define SN for any
positive integer N, so N is the positive integer, SN is the, by definition the set of bijections of the
set 1, 2, 3 up to N. And recall the exercise I gave in the first video, if you compose two bijections
you get another bijection, so that exercise as that SN is closed under composition, just like S3
was closed under composition, so SN is closed under composition, it has an identity element
namely, the identity function, remember this is the function which sends 1 to 1, 1 to 1, 2 to 2, 3
to 3, and so on N to N, it is the identity function because when you compose any function with
this function you get that function back.

Similarly every element, okay this requires a little bit of thinking, but we can prove this as every
element has an inverse, meaning if you give me any bijection of SN, there is another, sorry, any
bijection of 1, 2, 3 up to N, you can just construct another bijection such that composition gives
you identity. For example if one bijection sends 1 to 2, you can simply send 2 to 1 under a new
bijection, so the composition will send 1 to 1, so every element has an inverse and composition
as always is associative, so in other words SN under composition is a group, okay, SN is a group
under composition. We are going to spend in one of the future weeks, a lot of time on
understanding the group SN, it is called, it has a name it’s one of the important groups in group
theory, it is called the “symmetric group on N letters”, it is called the symmetric groups on N
letters, letters being 1, 2, 3, up to N, so it is a group on those letters, so it is called the symmetric

14
groups on N letters and as I said a very important group in the theory of abstract groups, and we
will study this later in more detail, okay.

So one more example that I want to discuss, okay, so just to recap a little bit we have several
examples of groups now, we have groups of numbers like Z, Q, R, C, positive rationals, nonzero
rationals, positive reals, nonzero reals, nonzero complexes, under either multiplication or
addition

groups, similarly we have symmetric groups, we studied S3 in the previous video in detail, now I
define SN in general that’s a group, we also talked about group of rotations of an equilateral
triangle.

And another group that I want to discuss this is also an important group for us, we will refer to
this again in future, let’s look at, I’m going to give you a set of complex numbers, so fix a
positive integer, positive integer N, okay, so N is a positive integer, and I’m going to define a
complex number theta N to be cosine + I sine 2 pi/n, okay, so complex numbers are of the form
A+IB, where A and B are real numbers, so here I’m taking cosine 2 pi/n + I times sine 2 pi/n, so
if you know little bit about complex numbers, it’s not important in general for the course but in
this example you need to know this, if you have a complex number of this type when you take
the nth power of this, in other words I’m taking cosine 2 pi/n + I sine 2 pi/n, so I’m taking the
whole power N, this by the properties of complex numbers simply happens to be cosine 2 pi + I
sine 2 pi, so what happens is, you multiply N with this 2 pi/n, with this 2 pi/n, but cosine 2 pi is
1, sine 2 pi is 0, so this is 1, okay. So theta N power N is 1, so we say that theta N is a “primitive
nth root of unity”, okay, so it’s a nth root of unity because theta N power N is 1, and it’s
primitive because theta N power M is not equal to 1, if M is positive number less than N, so it’s
not an Mth root of 1 for M less than N, it’s the smallest positive integer such that theta, a power
of theta is 1 is N, okay, so this is what makes it primitive. Again this is, if you do not know
complex numbers you can disregard this example, but now let me define G to be 1 theta N, theta
N squared, okay, here powers are, so I’m here the operation is, is simply multiplication of
complex numbers, okay, so I’m only, when I write theta N squared, I mean theta N times and
theta N, theta N power N -1 is theta N power, theta N multiplied with itself N-1 times, and
remember when I do theta N power N I get 1 back, so this is what I get, so this I claim under
multiplication is a group.

Let’s spend two minutes on why this is the case? Why is it a group? Because is it closed under
multiplication, if I multiply theta, what are the elements of this? Theta N power i is an element
times, theta N power i that’s another element, what is theta N power i times, theta N power j, it is

15
theta N power i+j, but because theta N power N is 1, theta N time power i+j will be equal to one
of these, because I mean it’s easiest if I say by example, if theta N power N-1 times theta N
squared is theta N times 2N, sorry, N+1 but this is theta N times N, power N times theta N which
is theta N, okay, so this is again in the group, in the set G, so it is closed? Yes, identity? Yes, 1 is
there, right that’s identity, inverse? So I’m asking again, remember what is a group? It’s closed,
binary operation is closed that we checked here, quickly checked, identity is there, yes,
multiplication is certainly associative so again that we don’t need to check separately, is there an
inverse? Yes again, because what is the inverse of theta N power i? That is simply theta N power
N-i because if I do theta N power i, times theta N power N-i I get theta N power N which is 1,
okay, so inverse also exists, so this is a group.

Okay, so this group G, the group G being what I define earlier in the previous page, and X is the
multiplication is a group, this cross is a multiplication is a group, is a group it is called, it’s an
important group, it is called the group of Nth roots of unity, it consist of nth roots of unity,
remember that every element of this group is an Nth root of unity, because if you do theta N
power i power N, then this is certainly equal to theta N power N power i, because this is just iN
and I can pull out i, and this is 1 power i, and this is 1, so everything here is an Nth root of unity,
however everything here is not primitive Nth root of unity, so it’s called the group of Nth roots
of unity.

A familiar example that you all know is, if you take N = 4, remember theta 4 is cosine 2 pi/4 + I
sine 2 pi/4, okay, so my notation I realize now is a bit confusing, here I refers to the square root
of -1, here i is just an index, so what is cosine 2 pi/4 that is just 0, that’s cosine pi/2 so this is
simply I, so theta 4 is a primitive, 4th root of unity which is I, so what is G? So I should really
call these groups GNs, so I should maybe call this GN, so G4 is 1, theta 4 which is I, theta 4
squared, what is I squared? I squared is -1, theta 4 cubed which is –I, and that’s all, so this is
simply {1, I – 1, - I}, so this is the group of 4th roots of unity.

And one more point that I’ll make which is useful sometimes to keep in mind is all the roots of
unity for any N are on the unit circle, so if you look at the unit circle this is 1, that’s a first root of
unity, this is -1 you have I here and –I here, in more general you have cosine 2 pi/N + I sine 2
pi/N, that’s this name, okay, and this angle is 2 pi/N, okay, so these are elements on the unit
circle, so this is another example, so again in this video also I’m basically trying to give you
examples of the groups, we have various groups now, we have groups of numbers, we have
groups of bijections of a given set, and now I have a group of Nth roots of unity.

16
One more important example is the group, groups of matrices, so groups of matrices are good
examples of groups in future for us, and I’m going to spend just a couple of minutes describing
them, and then as and when needed we will discuss more of these, so if you know what groups
are, matrices are, let’s say you fix two positive integers M and N, and we consider M x N
matrices, we have to specify where entries come from, so let’s just take R, so these are, matrices
where entries are real numbers, okay, so the set I think I called them M, N and so let’s just take
the notation is M m x n (R) is the set of all m x n real matrices, okay, so these are, for example
this is M rows and N columns, so we can there is an addition here, right, so you can add matrices
entry wise, so M m x n (R) is a group under addition, this is easy to check, for example if you
take, let’s say 2 x 3 matrices, so you have A11, A12, A13, A21, A22, A23, you can always add
two matrices B11, B12, B13, B21, B22, B23, there is no, the fact that it’s a matrix is irrelevant
here, it’s just about arranging certain real numbers, so you just add them component wise, okay,
so it’s just the same. Okay, so because addition under reals is a group you can quickly check that,
this is also a group the identity element for example will simply be 0, 0, 0, 0, 0, 0, so there is no
surprise, this is simply several copies of R, so that’s not surprising.

On the other hand you can also ask matrices form a group under multiplication? that also you can
ask, because there is a way to multiply matrices, immediately you have to impose certain
restrictions, because you can’t multiply a 2 x 3 matrix with a 2 x 3 matrix you can’t multiply,
right. In order to multiply 2 matrices the number of columns in the first one must equal the
number of rows in the second one, so restrict our attention to square matrices, remember that in
terms of the language that we are developing, multiplication is not a binary operation on the set
of matrices, if I don’t specify that they are square matrices, because I cannot combine 2 x 3
matrices under multiplication, but now I overcome that difficulty by considering square matrices,
here at least we can multiply and we do get a binary operation, so we do have a, because any two
square matrices can be multiplied to get another square matrix. However this is not a group
because not every matrix has an inverse. If you have studied matrix theory earlier you will see
that there are many matrices which do not admit inverses under multiplication, in order to do that
the matrix has to be what we call a an invertible matrix, so if you look MN(R) to be all N x N
matrices, over R, if you do that, we agree I hope, that MN(R) is not a group under multiplication
of matrices, because even though it is a binary operation, in other words multiplication is closed,
not every matrix has an inverse, but if you look at, so the reason is let me just write it here
because inverses do not exist in general, okay, so that is the reason why it’s not a group. But we
will restrict our set of matrices in order for inverses to exist, so now define, it’s usually denoted
GLN(R), to be the set of invertible N x N matrices over R, okay, so we have artificially restricted
our attention to only those which admit inverses, then GLN(R) is a group, okay, so again this
requires a small verification which I’ll leave for you to do, but I’ll just run through what you
required to do. We need to check that multiplication is a binary operation, certainly it is, if you
multiply 2 invertible matrices you do get another invertible matrix. Is there an identity element in

17
this? Yes, there is because the identity matrix is invertible. Are there inverses? Yes, because we
have restricted our attention to only those matrices which admit inverses, so every element in
GLN(R) has, this is GLN(R), every element in GLN(R) has an inverse and matrix multiplication
is associative that you have studied in some matrix theory course, so GLN(R) has all the required
properties of a group, so it’s a group, okay. This is an important group for us, so I’ll end the
video now, but to recall what we have done today in this video is first we started with the
definition of a group, and then have seen a series of examples of groups which will be very
important to understand in future when we study properties of groups. So in my next video I’m
going to quickly recall all the groups that we have learned and then study some properties of
groups.

18
 NPTEL

NPTEL ONLINE COURSE

Introduction to Abstract

Group Theory

MODULE – 01

Lecture – 04- “Basic properties of groups and

Multiplication tables”

PROF.KRISHNA HANUMANTHU

CHENNAI MATHEMATICAL INSTITUTE

So far we have in the previous videos we have studied several examples of groups
and we formally saw the definition of a group, so let’s continue this, we will study
in this video some properties of groups and some important types of groups, okay.
So let’s recall, I’ll spend 2 minutes of recalling, so if G is a group remember in the
other video I said G is a group under a specific operation, so let’s say under an
operation star, so that means it is closed under the star, in other words G x G to G
there is a binary operation, there exists an identity element, every element has an
inverse, and finally star is associative. The important examples that we have
looked at, I will not explain everything but ,Z, R, Q, C under addition Q star which
is non-zero rational numbers, R star which is nonzero reals, and nonzero complex
numbers under multiplication, positive reals and positive rationals under
multiplication, we also looked at SN bijections okay, and I think we looked at roots
of, nth roots of 1, unity, we looked at rotational symmetries of an equilateral
triangle, and we also looked at some matrix groups, so this is just to recall what we
have done so far.

So next I want to study some examples some properties and different types of
groups, so the first I want to, type I want to define is a group G, let’s say star is
called “abelian”, it’s also sometimes called a “commutative group”, but we will

19
most of the time use the word abelian if a star b is equal to b star a for all a, b in G.
So remember that a group is a set along with a binary operation and we say that it’s
abelian if it does not matter in which order we multiply or compose the any given
two elements, so and if you recall when I first looked at some examples of groups
and I looked at Z in particular, I noted that this property holds in Z, but I
commented there that this is not a critical property for a group, so we don’t ask for
it and we just give a new name to groups which have that property, so obviously
the examples will include Z, because the addition of integers is abelian, also
similarly Q, R, C, Q star, R star, C star, Q positive reals, positive rationals, positive
reals are all abelian, okay, because I am not emphasizing the operation here,
because it’s understood from the earlier videos, in Z, Q, R, C the operation is
addition, it is abelian. Q star, R star, C star, Q+, R+ there are all multiplication,
they’re abelian, so this is easy, usual multiplication and addition of numbers is
abelian, but not everything is abelian as we saw, SN is not abelian. We specifically
saw that S3 was not abelian in the earlier video where we looked at S3 in specific
detail and concluded that it’s not abelian, so actually I should write for N at least 3,
if you look at and I’ll give this as an exercise. Remember SN is defined for any
positive integer. So S1 and S2 are abelian, S1 is just a bijections of single element
set, so S1 itself will be a single element group, S2 is a bijections of a two element
set, so there will be two such bijections and it will be abelian, okay, so this is an
important class of groups, they are called abelian groups if the group operation is
commutative, meaning a star b is same as b star a, we also use the word, we mostly
use the word abelian, okay.

And also we have another natural definition, a group G is called finite, if the
number of elements, very natural definition, so G is a finite group if, it has finitely
many elements, okay, so as you can see here I’ve already omitted to write the
binary operation because it’s not required for the definition, it’s a purely set-
theoretic notion, a group is finite if as a set it is finite. So again examples: SN is
finite, okay, as an exercise you can do this, SN has, this is true for all N greater
than or equal to 1, SN has N factorial elements, and of course Z, Q and so on the
other list of groups, remember Z, Q, R, C, Q star, R star and so on are not finite
groups, okay.

20
A related definition: if G is a finite group then the order of G is defined to be the
number of elements of G, okay, it is denoted by this symbol, okay, so G with two
vertical bars is by definition the order of G, so for example, order of SN, by the
exercise I mentioned earlier is N factorial, okay, so we only usually talk about
order for finite groups, because otherwise number of elements is infinite, okay.

So next, I want to talk about, so important distinction between groups is between

finite groups and infinite groups, and we have examples of both kinds of groups in
our earlier examples. So now some properties I want to discuss of groups, these are
very general things, I will check some of them, leave the others for you to verify.
One property that comes in very handy is called cancellation property. What do I
mean by this? Okay, so let’s say G is a group, okay, let’s say I have 3 elements in
it, so a, b, c are in G. Cancellation as what to the name suggest, so I can cancel
elements in the following sense, so suppose that I have ab = ac, so here I’m writing
a times b, or a star b really in general, but just for the sake of convenience of
notation I don’t put star, I really mean, but as you will agree it’s easier to write
without star so I’ll just write ab = ac, but when I’m talking generally about a group
I will suppress this star symbol so and I write ab= ac, okay, so then so I have ab, a
star b, or a times b I use all this words, just its easiest to say ab = ac, so if ab = ac
then the cancellation properties says we have b = c, so ab = ac gives b = c, in other
words that is we can cancel a, so we are used to doing this in school, right, if we
have ab = ac we can cancel, this is okay, this is what I’m doing, but in an abstract
group this is a valid operation, this is what I want to now emphasize, why is this?
So let’s do this, first time you are doing, seeing this perhaps, so let’s check this
carefully step by step, so what are we given? We have given, we have been ab =
ac, so I’m going to use in the proof now, only properties of groups that are
contained in the definition, I do not want to use any properties of numbers or
functions or rotations, any kind of complex numbers, anything that I know before I
do not want to use, I only want to use group axioms or group properties, this is
what the subject of abstract group theory does, so I have, this is an equation in G.

Now I can multiply, so again let me remind you I use the word multiply to simply
mean, I apply the binary operation by a inverse, and a inverse also let me remind
recall a inverse is the notation for inverse of a, so if you have two elements that are

21
equal in a group by multiplying by a inverse they will remain equal, okay, so if ab
= ac, a inverse times ab = a inverse ac, that is just a set theoretic property, if you
have two elements and you multiply by the same element they will remain equal,
this is the meaning of binary operation. But now I’m going to use the associativity
of my group operation, so this is using associativity, so I have used the property
that we have associativity and I have earlier also used the property that I have
inverses, so remember that, that’s an axiom of a group. In order to even talk about
an inverse I need to have it in the group, which exists because it’s a group,
similarly I have used associativity at this point, but now because a inverse is
inverse of a, this is simply e times b, this is because a inverse a is identity, this is
the definition of inverse, and what is the definition of identity? e times b = b, this is
because e is the identity.

So in other words we have used all the properties, right, we have used
associativity, that we have inverses, that we have identity element, and of course
we have a binary operation has been used throughout, so please remember what I
have just, I wanted to prove, so you have ab = ac implies b = c. This should recall
for you something that you have done in school or for a long time, when you are
solving this should remind you, solving linear equations, okay, so you have 2X + 3
= 5X + 9 something like this, so here if you want to do this, if you solve this what
do you do? You can subtract 2X and subtract 9 so this will give you 3X = - 6
implies X = -2, so this we are used to doing in some sense without thinking a lot
whether it is valid or not, it’s not always valid, and in order to make it valid we
have define proper mathematical structures, and a group does that, a group allows
you a systematic correct setting where you can perform such operations, many of
the things adding, subtracting, multiplying, cancelling, all these are valid
operations in a group, so this is just a side remark, don’t think too much about it,
I’m just trying to justify why cancellation property is something that we are used to
dealing with, and in the new setting of an abstract group we have it.

Once you have a cancellation property we can do some more things. So a group
remember, a group always contains the identity element, I keep saying, in all my
videos I’ve been saying that existence of the identity element is an important part

22
of definition of a group, but I’ve been sloppy if you noticed in earlier videos,
whether sometimes I say it has an identity or the identity, but now I will justify
everything by saying a group has a unique identity element, namely a group cannot
admit two distinct identity elements, and hence we are allowed to say the identity
element of a group, why is this? Why does it have a unique identity? So suppose
not, so suppose so why? We have to prove this, right, anything we write we have to
prove, so suppose that G has two identities, let’s say e and e prime, you note that
we have been using the letter e to denote identity element of a group, so e and e
prime let’s say serve as the identity, and at this point let’s recall also the definition
of the identity element, what is an identity element? It’s an element which has the
property that when you multiply any element of the group with it, you have to get
the element back.

So for example when you multiply e with e you get e, this is the definition of the
identity element, so e is the identity element, small e is some other element, I mean
it’s some element of the group, so e time e is e, but what is e times e prime? e
prime is also the identity element, an identity element of the group, e is an element
of the group so e times e prime must be e, because e prime is an identity element,
this is because e prime is the identity element, e prime is the identity element
means g times e prime is equal to g, for every g in the group G, I’m applying it to
e, so e times e prime is e, but then e times e is also e, so e times e is equal to e
times e prime, but now cancellation property says, remember this is exactly the
setting of the cancellation property ee = ee prime that means e = e prime as we
wanted, so if you start with two identities they are equal, so a group has exactly
one identity, it must have one identity, it cannot have two different identities, also
so this is one property that immediately follow from cancellation property.

Another property, any element of G has a unique inverse, okay, again if you go
back and see the videos earlier, I’ve been sloppy sometime I say, let g be the
inverse of g, let g inverse be the inverse of g or let it be a inverse of g, I’ve been,
maybe inconsistent in my usage, but again I will prove that inverse has to be
unique, so let say g is an element of, all the arguments here are very standard
arguments, but they are critical arguments to understand group theory, how to

23
work with abstract groups, in order to understand and familiarize yourself with
these things, you have to understand these arguments, they are simple but if you
are seeing them for the first time they will require some getting used to, so let’s say
g is, a small g is an element of capital G, so suppose g has two inverses, say g1 and
g2, remember my claim is that every element has a unique inverse, certainly it has
some inverse because that’s the definition of a group, every element has an inverse,
but suppose it has two inverses, so then we now that g and g1 is equal to the
identity element, by definition of an inverse, but so as g times g2 because g2 is also
an inverse of g, but again cancellation property gives you, forget this, you apply
cancellation property to this, it gives you g1 = g2, so again very simple, but you
conclude that, cancellation property immediately gives you inverse has to be
unique.

One more such thing I want to say, if you take, let’s say g1, g2 are elements in a
group, so suppose that g1g2 is e, then automatically g2g1 is e, okay, so remember
that I’m working with an arbitrary group G, it’s not necessarily abelian, I cannot in
general switch, so note in general G is not abelian, so it is not true that g1 times g2
= g2 times g1 for two elements g1, g2 of G. It is certainly not true that g1g2 = g2g1
for two elements in a group in general. However, if g1g2 = e it must automatically
mean that g2g1 = e, so in other words if g2 is inverse of g1 they commute, and this
is again an easy application of the cancellation property. Let’s try g1g2 = e, then,
so then what do we do? So if you multiply by, so I want to conclude that g2 = e,
g2g1 = e, let me now continue so g1g2 = e, so by multiplying both sides by g1
inverse, what do I get? g1 inverse, okay, so I don’t need this, so I have this, so I’m
just multiplying this equality by g1 inverse on both sides which I can do. Now
apply the associativity, this is very similar to the work we have done in proving
cancellation property so we have g1, remember g1 inverse e is equal to g1 inverse,
because e is the identity element, this means because g1 inverse g1 is e, and finally
this means g2 is g1 inverse, note that that means if g1 times g2 is equal to e using
group axioms so we are able to conclude g2 is g1 inverse, and earlier in the
previous slide I have shown that, inverse of an element is unique, so g2 is the
unique inverse of g1, so that means now let’s see what we want to prove, g2 times
g1 to prove, the whole thing here is to prove if g1 g2 = e, then g2 g1 is also equal
to e, so why is this true, but because we have already concluded g2 is g1 inverse,

24
so g2 g1 is g1 inverse g1, but g1 inverse g1 is, by definition of the inverse, identity,
so if g1 g2 is identity, g2 g1 is equal to identity. In order to check something is the
identity element it’s enough to check the, something is the inverse, it’s enough to
check that product in one direction is identity.

And similarly one more property, maybe I’ll leave these as exercises for you, very
similar to the work that we have done in these things, so if g is an element of G,
then g inverse whole inverse is g, so this is exercise one for you, this means that
inverse of g inverse is g which is actually if you just write out what it means is
obvious. Similarly if g and h are elements of a group G then the inverse of gh is h
inverse times g inverse, so these are also easy exercises that I encourage you to do,
just they will follow from the definition of inverses and group axioms, okay.

So and some more notation now I want to introduce in order to talk about
multiplication tables, so if G is a group one can write down multiplication tables
which are a compact way of describing the entire group, so and these make sense
really only for finite groups, so let G be a finite group. Recall that a finite group is
a group which has only finitely many elements. So for example S3 or group of, so
a group of nth root of unity, these are all finite groups, so multiplication table as I
said is a table which completely describes the group, so for example if you take the
group of fourth roots of unity, if you recall from earlier today, earlier in the videos
I defined these group, which is the group of fourth roots of unity, so every element
in this group is a fourth root of unity, so here of course i is the square root of -1, so
i is the complex

square root of -1, so if this group the multiplication table of, by multiplication table
of this group, I mean the following, so I’m going to draw a table like this, I’ll have
one row for every element of the group, and similarly one row, one column for
every element of the group, okay, so I’m going to draw a grid in some sense, so
multiplication table, so clearly this only is possible, if the group has only finitely
many elements, if it has infinitely many elements we cannot just contain the
information in a table like this.

25
So in any position for example if this position, I have i in the row and -1 in the
column I’ll write down the product, so in this case it’s –i, so for example let’s just
do one, one times anything is itself, so this is just this, -1 is -1, 1 – i and i, here I
have i, i times -1 is –i, i times i is i squared which is -1, i times –i is –i squared
which is 1, -i times 1 is –i, -i times -1 is i, -i times i is 1, -i times –i is -1, okay, so
this is the multiplication table of this group of fourth roots of unity. And if you just
stare at this table for a minute, and if you forget the first row and the first column
only focus on the interior part of this table, so these are just labels, okay, forget
those, each row of the table is just listing the elements of the group, no element can
repeat here, and no element can miss from this list, for example the group is 1, -1,
i, –i, here you have that in different orders you have listed all these elements.

Similarly each column contains all the group elements in some order, for example
this column here i, – i, -1, 1 this is really a property of cancellation that’s, you can
do in a group, okay, so because you can cancel no two elements in any row or
column can repeat themselves, okay. And as an exercise and this is again a very
good exercise for you to familiarize yourself with calculations in a group, write
down the, remember S3 from earlier video, S3 was the group of bijections of a 3
element set, so if you recall I have used this notation, okay, there were three such
bijections if you go back to the beginning first video, I’ve completely described
each element here F1 was the identity bijection, F2 is the bijection which fixes 1
and interchanges two elements, so I don’t remember exactly what it is, but go back
to it and write down the multiplication table, so along these lines that we have done
for this group, it will have 6 rows and 6 columns and you have to fill in each, each
spot in the table and as a way to check your answer, make sure that it, the table that
you come up with that the end of the work has a property that this entire group is
listed in each row and each column, if some element is repeated then you have a
made a mistake, if some element is missing you have made a mistake, so make
sure that you verify that each element appears in each row and each column
exactly ones.

So I’ll stop with this today, in this video please make sure that you do the exercise
that I have assigned and when we meet next time we are going to study subgroups
of a group and look at more properties of groups.

26
 NPTEL

NPTEL ONLINE COURSE

Introduction to Abstract

Group Theory

MODULE – 01

Lecture – 05- “Problems 1”

PROF.KRISHNA HANUMANTHU

CHENNAI MATHEMATICAL INSTITUTE

Last videos we have looked at definition of a group, we looked at various

examples, some basic properties of groups. So let me start today this video by
working out some examples and exercises on groups, then we will study some
more notions of groups, okay. So let start with a question, so I’ll do a few exercises
now and I’ll work them out in detail so that you can do more similar examples.

So let’s say I ask whether the following are groups, okay. So first let’s say the
underlying set is Z, we have seen in the previous videos that Z is a group under
addition, but here I am giving a different binary operation, let’s say I define, this is
the definition, so I’ll define a star b to be a – b, so certainly this is a binary
operation because if I take two integers apply star which in this problem is just
taking the difference you get an integer, for example if you do 5 – 3 you get 2, so
similarly 3 – 5 is -2, so it is a binary operation, so far so good there is also an
identity element, right, because 0 is the identity element, if you subtract 0 you get
the element back, but actually I have to be careful, right, is this true really, is 0 the
identity element? Certainly if you do a – 0, so a star 0 is a – 0 which is a, that is
okay, but remember that we want 0 star a to be also a, but 0 star a is 0 – a which is
– a, so actually 0 is not the identity element and you can check that, so 0 is not
identity.

27
You can in fact check that there is no identity, that is because really 0 is the only
possible identity because when you subtract 0 from a you get a, if you subtract
anything else you don’t get a, and you can also check that star is also not
associative, okay, so I will simply write here, for example if you do 5 – (3 – 2),
that is 5 – 1 that’s 4, whereas (5 – 3) – 2 is 2 – 2 which is 0, okay, so it’s not
associative, so this is not a group, so this is not a group, okay, so this should
suggest to you that operation is very important, right, subtraction is a very natural
operation but under it the integers do not form a group.

Okay, let me look at a different example now, let’s look at the same set, let’s look
at the same set but now I define a new operation, again the set is set of integers, but
the operation is the following, so I do a star b is defined to be a + b + ab, okay, so
this is again the definition of an operation, so for example if you do 3 star 5 you get
3 + 5 + 15, so that’s 23, so this is certainly a binary operation, because if you
perform this operation on integers you to get an integer, so this is a binary
operation.

Is there an identity element? So if you do a star 0 you get a + 0 + 0 which is a, and

if you do 0 star a you get 0 + a + 0 which is a, so this is true for all a in Z, so 0 is
the identity element so that’s good, so we have an identity element it’s a closed
operation. So is it associative? How do you check associativity? We want to do a
star b star c what is this? This is, first I’m doing a star b so this is a + b + ab star c,
this is, remember that when I, star I take the first one plus the second one plus the
product of the two, so here the first element is a + b + ab, second element is c plus
the product, so that is ac + bc + abc so that is the element a star b in bracket then
star c.

On the other hand if you do a star (b star c) this is a star b + c + bc, okay, now
again remember star means we take the sum of the two terms and then add the
product of the two terms to this, first two terms so you get a + b + c + bc + the
product which is ab + ac + abc, so now if you actually look at this two these are
equal, because a is there, b is there, c is there, ab, ac, bc are there, abc is there, so

28
these are equal. And so star is actually associative also, okay, so it has three of the
four properties that we want, but we need to check now for inverse, so what would
be an inverse of this? So we want, and we have already noted that the identity
element is 0, so we want, given an element a in integers, we want, what do we
want? We want b which is another integer such that a star b is 0 because that’s the
definition of inverse, and remember that b star a is also a so that is okay, if b has
this property it will automatically have this property.

Okay, but what is a star b, that means a + b + ab must be 0, okay, so we know a,

we have to solve for b, so let’s try to solve for b, so this will be a + b times (1 + a)
= 0, so we can try to write it like this b times 1 + a = - a, so b should be – a divided
by 1 + a, okay, this is what b should be if a star b is 0, but now there are problems
here, first of all this is not an integer, if a is an integer this is not an integer, in
general right because you are dividing by this, so for example if a is 2, b should be
– 2 divided by 3, this is not in Z, so star does not have, does not admit inverses,
okay, so (Z, star) is, we can immediately conclude that it is not a group.

Note that it has 3 properties, that’s okay, binary operation identity associative, but
it doesn’t admit inverses so it’s not a group, okay. But let’s look at bit more closely
at this operation, why

did inverses not exist here, because we have to consider rational numbers, so how
about enlarging our set, so how about the same operation star but on let’s say Q,
the set of rational numbers? So here in the previous page we saw that the inverse of
2 under this operation should be – 2/3 which is not an integer but it is a rational
number, so it looks like star must admit inverses in rational numbers, but
remember what is the inverse of a? It should be, if you recall, it should be – a
divided by a + 1, this is okay, it is in rational numbers but there is a problem, it is
not defined when a is -1, because when a is -1 the denominator is 0 here, so this is
not a well-defined number, so even if you take the larger set of rational numbers
the star operation will not have inverse for -1, so it will have inverses for every
other element, because 1 + a will not be 0, so – a divided by 1 + a is a well-defined
rational number, so a = -1 is only problem.

29
So now we set, let’s Q minus, okay, so in other words I’m removing – 1 from this
set, so I am looking at the set of rational numbers, different from – 1, so let’s say G
is this, if G is this so I should not write, G is the collection of rational numbers
different from – 1, then what we have verified says that G star is a group, recall
that star is again as before it is a + b times a + ab, so under this G star is a group, if
you eliminate – 1 from Q you can make it a group, okay, so this

example is a good example to keep in mind, because we started with Z under this
operation, we started with Z under this operation we saw that it’s a binary
operation it has identity element, it is associative, however it doesn’t have inverses
in Z, so if you enlarge your set and consider all rational numbers then it has
inverses except when we are looking at – 1, so we have removed it from the set,
and we’ve only looked at rational numbers different from – 1, then G star is a
group, okay. So sometimes we have to add more elements, sometimes we have to
delete some elements in order to obtain a group.

Let’s look at another example, so I don’t recall the numbering here, so maybe this
is the third example, so let’s look at the set of rational numbers which have the
property as I will write now, so rational numbers can always be written like this,
okay, they are ratios of integers, so let’s say a and b are co-prime, co-prime means
they have no common factors, we say that a and b are co-prime, so I’m going to
quickly define in case you are not familiar with this word, if they have no common
factors, okay, so for example 6 and 10 are not co-prime, because 2 divides them.
On the other hand 2 divides 6, 2 divides 5 so they have a common factor, on the
other hand 6 and 9, actually 6 and 9 are also not co-prime, but if you take 6 and
let’s say 25, only factors of 25 are the 1, 5, and 25, factors of 6 are 1, 2, 3 and 6, so
there is no common factor so these are co-prime.

When you write a rational number you can always cancel all the common factors
so you can write it in the reduced form, so I take reduced form, so a, b are co-prime
and suppose that 5 divides b, this is my set, so the set that I’m considering is set of
rational numbers in the reduced form where the denominator is divisible by 5, so
for example if you have 1/5 is in G, remember but 1/4 is not in G, remember that

30
you have to consider the reduced form, for example 1/5 is also same as 2/10, sorry
so 1/4 is also same as 5/20, right, and you can look at this and say 5 divides 20, so
5/20 belongs to G that’s not correct, I must first cancel all common factors and
then ask whether 5 divides the denominator, okay, so 1/4 is not in G.

Is this a group, is G a group under addition of rational numbers? So I can add

rational numbers in fact Q under addition is a group, I’m asking for a subset of that
under the same operation is it a group, so what is the, so here you can check that
various properties fail here, for example even the closure is not true, for example if
you take 2/5 and 3/5 to it, remember 2/5 is in G, 3/5 is in G, 2/5 and 3/5 are in G
because they are in the reduced form, and 5 divides the denominator, but what is
2/5 + 3/5 that is 5/5 which is not in reduced form, so you have to cancel and you
get 1/1, it is now in the reduced form, but 1/1 is it in G? you have to ask yourself
whether 5 divides the denominator, it does not, so 1/1 is not in G, so this is not a
group because it’s not even a binary operation, so it’s not a group because G is not
closed under addition, so this is not a group, so is this clear? So G, as defined here
is not a group because it’s not closed under addition.

Now let me modify this, let me take G prime, I use this G dash to denote a
different set now, this is all rational numbers, again a and b are co-prime so the
rational number is in reduced form, but unlike in the previous example now I
demand that 5 does not divide, so by the way when I write this symbol I should
have remembered when I write 5 this symbol, this means 5 divides b, so b is
divisible by 5.

And now let say 5 does not divide b, so it is very similar to the previous example
but it’s very different also, because now I want the denominator to be not divisible
by b, for example here 1/2 is in G prime, but 1/2, 1/5 is not in G prime, 5 does not
divide 2, so 1/2 is in G prime, 5 divides 5 so 1/5 is not in G prime.

31
Is G prime a group under addition? So now the same question, the previous
example also I asked is G prime a group under addition, is G a group under
addition, and we saw that it doesn’t even satisfy the first property, it’s not a binary
operation, addition is not a binary operation so I conclude that G here was not a
group.

Now I have modified G, I’ve looked at G prime and now I’m asking is it a group
under addition, so let’s ask, let’s take a/b and c/d are in G prime, so let them be in
G prime, so in particular remember that they are in reduced form, and 5 does not
divide b, 5 does not divide d, what is their sum? So if you sum them it is a/b + c/d
and this is simply ad + bc divided by bd.

Now the question is, is this in G prime, we’ve started with 2 elements of G prime,
we have taken their sum and seeing, asking whether the sum is in G prime, the
previous example the sum is not in G, sum of two things in G is not in G, in this
case, is it true?

Now the point is first of all this may not be in reduced form, but that doesn’t matter
because note that 5 does not divide, remember by definition 5 does not divide a,
and 5 does not divide b also that is because, that is because a/b is in G prime, and
a/b is in reduced form, so 5 does not divide b, c/d is also in G prime so by
definition of G prime 5 does not divide d.

Now a property of prime numbers says that if 5 does not divide a, and 5 does not
divide b, 5 does not, sorry, I should not write a here, remember 5 does not divide b
is what I should write. 5 does not divide the denominator, 5 does not divide b, 5
does not divide d in the second example, so 5 does not divide bd, because one
definition of prime number is that if 5 divides bd the product, then it must divide
either b or d, so that’s not the case, it does not divide b, does not divide d. So if 5
does not divide bd, now let’s look at this, let’s look at this rational number, while it
may be the case that this is not in reduced form and you might have to cancel some
factors, but 5 does not divide bd as it is already, so if you further cancel some
factors there will be fewer factors, so 5 will definitely not divide that new factor,
after cancelling whatever is the denominator 5 does not divide it also, so in the

32
reduced form after you find the reduced form of this you can conclude that this is
in G prime.

So let me repeat the argument again, why I am saying that this is in G prime,
because 5 does not divide b times d even after cancelling common factors of this
ratio whatever is the new denominator also will have the property that 5 does not
divide it, because in the new, after cancelling the new denominator will have fewer
factors than bd, bd does not have 5 as a factor so after cancelling some whatever
you get will continue to not have 5 as a factor, so this will be in G prime.

So in other words the upshot is that G prime is closed under addition which is the
first property that we want for a group, so addition is a binary operation on G
prime. So now let’s verify the other properties, for example what is the identity
element for addition? It must be 0, so you have to ask is 0 in G prime, it is, because
0 is 0/1 and 5 does not divide 1, so 0 is in G prime, this is okay.

What is the inverse of a/b? Inverse of a/b is - a/b, and if a/b is, if a/b is in G prime,
then 5 does not divide b, right, that is the definition, it’s in the reduced form and 5
does not divide the numerator, so then obviously – a/b will be also be in G prime
because denominator is the same, 5 does not divide b, so inverses exist and
addition is certainly associative, that is because addition is associative on the set of
rational numbers, so this is okay, inverses exist it is closed under addition there is
identity, so G prime is a group, we conclude that G prime is a group under
addition. So again this example is very nice to keep in mind because we have, in
earlier we have looked at G which is all rational numbers were 5 divides the
denominator it’s not a group, whereas if you modify the definitions slightly and
say 5 does not divide b, it becomes a group, okay, so this is an important group that
we will, it’s an example of what we will later call a subgroup of a group, so we
will come back to this later, but we will say that G prime is actually, I’ll write it
here just so that you keep in mind, I will refer to this later, G prime is a subgroup
of Q, okay.

33
So this is as of now it may not mean anything to you, but we will keep, we will
come back to this.

34
 NPTEL

NPTEL ONLINE COURSE

Introduction to Abstract

Group Theory

MODULE – 01

Lecture – 06- “Problems 2”

PROF.KRISHNA HANUMANTHU

CHENNAI MATHEMATICAL INSTITUTE

I want to do some more examples, exercises again to illustrate properties of groups,

so that you become comfortable with operating with groups.

So let me ask you a new problem, so here let’s say G is a finite group, in this
problem I’m working with a finite group, recall that in other words, remember
what are finite groups from my earlier videos, G has only finitely many elements,
so the problem is the following. So I’m giving you an exercise that I will solve in
detail, so G is a finite group, I want to show that, show that for every element of G
there exists a positive integer N such that A power N is identity, okay, so just a
piece of notation because maybe I’ve not clearly defined this in the previous
videos, when I write A power N will mean A star A star … star A, N times,
whenever I write A power N I always mean this, because this is a short cut, so star
is the operation, remember here we are working with an arbitrary finite group, and
you will have to get used to thinking like this, it’s not Z, it’s not Q, it’s not S3, it’s
not roots of unity, so it’s not any specific thing, it’s any group and the

problem should not use any specific properties of examples of groups, it must only
use properties that all groups have, so here G is a finite group and star let’s say is
the operation, it’s a shortcut, it’s an easy notation instead of writing A star A star A
star A always N times we will simply write A power N, so that is what I mean.

35
So now let’s come back to the problem, it’s asking you show that for any element
A of G there exists a positive integer N such that when you apply A to itself n
times you get the identity element, also remember E is the identity element, that is
my standard notation for a group, E will stand for identity element, generally when
I’m working with a general group E will stand for the identity element, so what is
the solution to this? So G is a finite group, so we’ll have to use that, so let’s do the
following, so consider, I’m going to consider the elements E, A, A squared, A
cubed, A power 4, A power 5, like that and I can take A power 100, A power 101
and so on, so what am I doing this here? So I’m starting with identity that is you
can think of identity as A power 0 always, that is again notation, A power 1 that
means just A, A squared means A star A, then A star A star A, A star A star A star
A, A star 5 times, 100 times, 101 times. Now these are all elements are, where are
these living? All these are elements of G, why is that? Because, why is this?
Because G is closed, I’ll write G is a group, star is a binary operation on G right,
so A is in G, remember A is in G that is starting point, A is in G so A squared is in
G, A cubed is in G, A to the 4th is in G, A to the 5th is in G, A to the 100 is in G,A
to the 101 is in G and so on, so these are all elements of G, but remember that G is
a finite group, we are going to use this very important hypothesis, G is a finite
group.

And here we have seemingly infinitely many elements, right, because you have A,
A squared, A cubed, A power 4, you can keep doing this, you can do A power one
thousand, A power ten thousand, A power one lakh and so on, but how do you
justify this? Now G is a finite group, and these are all elements in that finite group
and they are seemingly infinite but they cannot be infinite, so what is the
implication here? So for some positive integers N and M, we must have, right is it
clear, because this number, this elements A, A squared, A cubed are all different
that will be a infinite set contain in set G which is a finite set, which is impossible,
which is an absurd statement, so this cannot be an infinite set, which means there
must be repetitions, if this is not infinite set then some two elements in fact many
of them will collapse, so I am just saying that for some positive integers N and M
and of course I have to insist that N is not equal to M, otherwise A power N is

36
certainly equal to A power M, so for some positive integers N and M which are
different, we must have this.

Again, if not, if this does not happen then A power N is different for each N, and
this set A power N, the set consisting of all powers of A is an infinite set, living
inside a finite set giving you a contradiction, so A power N is equal to A power M
for two different positive integers. Because N is not equal to M we can assume
without loss of generality that N is strictly more than M we can assume that
because one of them is bigger than the other, we will simply say N is bigger than
M, then let’s now play with group properties, so we have A power N equals A
power M, so let’s multiply this with A power – M. Okay, now again what is A
power – M? If you are confused about this, I’ll make a remark here, what is A
power – M? It is just a short hand for, you take A power – M is by definition you
take the inverse of A and write it like this, okay, so it’s convenient to write this
element as A power – M, so this is just notation, so this is notation, instead of
writing A power -1 power M always so here what we are really doing is, so let me
just, because I’m doing this for the first time, I’ll just for clarity multiply both
sides, so in other words I have A power N, A power M, I get A power M, but now
let’s see what happens, so this is now let’s recall what is a definition of A power N,
so this is N times, A power – 1, so this is really again star, in general the operation
is star, so A power – 1 star, A power -1 M times, so that is the left hand side.

37
On the right hand side we have A power M, which is A times, A times A M times,
star A inverse, A inverse also M times, right, so this is just expanding these terms,
so but this is you can cancel this, what is the right hand side now? You have A
times A inverse which is identity so that will cancel, the previous one will cancel
with this and finally this will cancel, because this is exactly the same number, this
is M, this is also M, this is E, this is E, and here but we are assuming remember N
is strictly more than M, so we can’t cancel all of them, so we can cancel M of
them, so cancel M of them what will be left on the left hand side? We cancel M
copies of A on the left hand side using A inverse, so then what will be left with? Is
A star A, so this is, left hand side is E, right hand side is, after cancelling M copies
of A from here, how many will be left with? We will be left with N – 1, M copies,
okay, so that means A power N – M is E, okay.

Let’s go back and see what the problem asked us to do, we asked, problem asked
us to do, given an arbitrary element of the group there exists a positive integer N
such that A power N is E, did we get that? Yes, we got that, because N – M is a
positive integer which is a positive integer by the assumption that N is strictly
more than M, and we did get a positive integers such that A power that is E, so we
have solved the problem, so I hope this calculation here is clear to you, maybe just
a quick calculation I mean if you in case you’re confused about this, I just want to
do it with specific numbers, so let’s say N is 3, and M is 2, and just N is 5 and M is
3 let’s say in the previous example, so that means A power 5 is equal to A power 3,
I’m multiplying with if you go back and see I’m multiplying with A power -1
whole power, the smaller one which is M, so here I get A power 5 times, A power
– 1 power 3 is equal to A power 3, A power -1 power 3, so that means I have a
product of A 5 times so I’ll cancel 3 of those, so I get A squared and here I get E,
so this is just specific example so that you get understand this, so we have solved a
problem, we have solved the problem which asked in a finite set this happens.

Remember that we have crucially used the finiteness hypothesis, because we want
to say that this set is not finite, because it’s inside a finite group, it’s not finite and
hence there must be repetitions, so it must happen like this, okay.

38
And if you can see this is not going to be true, this is not true if G is not finite, for
example we take, let’s say G is Z and we take the element 1 and G, so Z is under
addition, okay, so I should, I should mention this because Z has, it’s just a set in
order to make it a group we will take the addition. If you take 1 in G remember the
notation that I’ve been using is A power N, that is just A star A star A, N times,
here the operation is addition so it’s a bit confusing to think of that power notation
in terms of addition, here spelling out exactly does there exist a positive integer N
such that 1 + 1 + … + 1 is the identity, remember A power N is E, so that means
we want A star

A star A, N times does there exist something like this, so 1 + 1 + 1, because the
addition is the operation here, and the identity element is 0, does there exist
something like this? Certainly not, because right, this is because 1 + 1 + 1…, N
times is actually N, but N can’t be 0, you’re asking for M to be positive, so N can’t
be 0, so it’s not true if G is not finite, this property is very specific to finite groups.

So just I’ll end to this video with one more example which is very similar to this,
so a different problem now. Let’s say G is a finite group again as before, show that
there exists a positive integer N such that A power N is E for all, okay, so if you
read the problem carefully, let G be a finite group, show that there exists a positive
integer N such that A power N is E for all A and G, what is the difference between
this and the previous problem? Here I am asking you show that for a given element
is A, there exists a positive integer N such that A power N equal to A, so this N has
to be chosen after I get A, so it can depend on A, in the new problem I’m asking
for an N which works for every element of G, so N in particular should not depend
on A, but this is now easy given that we have already solved the previous problem.
So solution, by the previous problem for any A in G there exists a positive integer,
let me now call it N sub A, because this positive integer from the previous problem
depends on A, so let’s call it N sub A such that A power N sub A is 1 or rather E,
okay, remember this is the content of the previous problem, there is a positive
integer NA such that A power NA is E.

39
Now we have one such positive integer for every element of A, every element A of
G, but again let’s use the fact that G is a finite group, so we have NA for every
small A in G, so now I simply define N to be maximum or rather let’s say product
NA, A and G, so this is like if G is A1 up to AR, G is a finite group right, so I’m
defining it to be NA1, NA2, NAR, for every element of the group there is an
associated positive integer, I’m taking the product of those positive integers, now I
claim that this N will have the required property.

We claim that this N works, in other words that is A power N is E for all A in G,
why? Why is this true? This is clear if you think about this for a minute, so N
remember is NA1.. NAR, right, so now if you take A power N, A is equal to one of
the AIs, so A power N is A power NA1.. NAR, okay, but let’s say A1 = A, just for
simplicity, okay, so actually let me, A1 power N, so I have A1 power this, right,
because N is equal to this, but if you now think about this it is, what is this? This is
A1 star A1 star A1, how many times, it is NA1, NA2, NAR times, okay, so you
can now combine this in NA1 times, okay. And then combine this again in NA1
times like that, so what I am saying is, so because this number is divisible by NA1
I can break up this huge product into smaller products, each one consisting of NA1
times, but remember A1 power NA1 is identity, that is how NA1 was chosen, so
this is identity, this is identity, this is identity, so the content of all this is that you
can use the usual exponential rules, that means this can be written as A times NA1
power NA2, NAR, okay.

Usual exponential rules that you’ve learned in school say that this is equal to A1
power, the product of these two, but A1 power NA1 is E, and E power anything E
is E, so this is the shortcut of the proof, okay, but the proof of this is this, because
I’m using the exponential rule as a, exponents as a notation, I can’t use exponential
rules that I have from integers and real numbers, and rational numbers, I have to
justify that, which this does. So maybe I’ll leave this as an exercise for you, if A
power N is identity and N divides M, then A power M is also identity, so this is an
exercise which I have essentially done here, but maybe not in great detail so I
would like you to all, I would like all of you to do this carefully, if A power N is
identity, and N divides M, A power M is also identity.

40
So now given this exercise if you accept this, the problem is solved, so we did get a
positive integer globally independent of elements of A such that A power N is
identity. So this I’ll end this video with this and in the next video I’ll do one more
problem in a lot of detail, and then we will continue our study of groups.

41
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
MODULE – 02
Lecture – 07 - “Problems 3”
PROF.KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

In this video I want to do one problem in detail which is very useful to understand
basic notions of group theory, because many of the properties that we have so far
learned are used in this problem, and if you understand this problem fully it is very
good and it will tell you that you’ve understood everything that has so far been
covered.

Okay, so the problem that I want to discuss in this video is the following. Show
that any group G of order less than or equal to 5 is abelian, so that’s the problem.
So I’m going to recall some terms here that I covered in earlier videos. What is the
order of a group? So recall that if G is a group and actually, if G is a finite group, G
is a finite group, the order of G, which is denoted by this symbol G within bars, is
the number of elements in G, okay, so order of G is the number of elements in G,
so in the problem the question is to show that every group which has order less
than or equal to 5, meaning any group which has 5 or fewer number of elements is
abelian. And what is abelian? G is abelian if AB = BA for all A, B in G. So we say
that A and B commute, so this symbol that AB = BA, in words means that A and B
commute with each other, which is to say that the order in which we multiply them
is irrelevant for the final answer, so G is abelian if every pair of elements in the
group commute with each other.

And this is the property that groups have and not every group is abelian that also I
want to recall before we get to solving this problem, if you recall the group S3,
symmetric group on 3 letters is the group of bijections of a 3 element set, which we
denoted by 1, 2, 3, and if you recall in the earlier video where I discussed S3 in
detail, we actually listed all the elements of S3 and understood some facts about
these elements, we saw that order of G was 6, and S3 was not abelian, okay. So if
you go back to an earlier video you will see that we have explicitly showed that
there exist two elements in S3 such that they don’t commute, okay, so that S3 is not
abelian and it has 6 elements, and the problem asks you to show that any group
which is smaller than that, meaning having 5 elements or 4 elements or 3 elements
or 2 elements or 1 element must be abelian, so the smallest group which is not

42
abelian is of order 6 which is S3.

So now let’s go ahead and solve the problem, so let’s solve it, so the basic idea of
the solution is the following, we are going to show essentially that, if a group has 5
elements or less there are just not too many elements, just not enough elements for
a pair to not commute with each other, okay remember that if a group is not abelian
it must be the case that there exists two elements that don’t commute with each
other, so maybe I should write this, so if G is not abelian because the definition
remember is that if G is abelian if, so the definition is that G is abelian if you give
me any pair of elements A, B in the group AB = BA, this must be true for all
elements A, B, so what is the contrapositive of this? If G is not abelian there exist
two elements, let’s say A, B, in G which don’t commute with each other, so such
that AB is not equal to BA. And the goal for us in this solution is to show that if a
group has 5 or less number of elements it cannot have just that’s not is, 5 is too
small a number to admit two element which have this property, which don’t
commute for each other, so we are going to consider 5 cases, we note that a group
is a nonempty set, because a group has to contain the identity element, that’s one of
the properties of a group, so group has at least one element. In our problem we
have to show that any group which has less than or equal to 5 elements is abelian,
so we will consider cases, so first case order of G is 1, if order is 1 that means there
is exactly one element in G and every group must contain the identity element so it
must be E, and this is certainly abelian, why? You can see this, this is very easy to
see it’s abelian because in order to be not abelian there must be two elements
which don’t commute with each other, but the G has only one element so there is
no question of having two elements which don’t commute with each other, so this
is okay, this is very easy.

Case 2, order of G is 2, so let’s enumerate these elements, so as I said earlier every

group contains E, so that must be the identity is one of the elements, okay. Now in
this case there is one more element which we will call A, so that means E and A are
the two distinct elements of this group, so note that in particular we have A is
different from E, right, because G has two elements one is E, the other we are
calling A, so certainly A is different from E.

Now if G is not abelian, again let me remind you if G is not abelian there exist two
elements which don’t commute with each other, in this case if G is not abelian that
means E and A do not commute with each other, but that’s certainly not the case, I
mean E and A do commute with each other, because every element in a group
commutes with identity, right, so recall that in general any element in a group
commutes with the identity element. So EA is, that is part of the definition of a

43
group that A, which are both individually equal to remember A, so G is abelian in
this case also, right, because the only two elements are there and they commute
with each other, so G is abelian, so in this case G is an abelian group, so we are
done with order 1, order 2, we have to still consider order 3, 4, 5.

So case 3, order is 3, by the same convention as in the previous case, I am going to

enumerate and give names to the elements E is one element that is the identity
element, let’s pick A as the second element, and let’s pick B as the third element.
So as before we have E, A, B are distinct elements. In other words A is different
from E, B is different from E, and A is different from B, G has 3 different elements
and I’m calling them E, A, B, E is the identity element, A and B are non-identity
elements.

Okay now again if G is not abelian it means that there exist 2 elements which do
not commute with each other, so only possibility in this case must be A and B,
because A and E commute and B and E commute, the question is do A and B
commute? Do A and B commute? So let’s do the following so let’s look at what
can be AB, so what can be AB? By which I mean, remember G is a group, and I
have given already names to its 3 elements, G is a group of order 3 and I have
already given names to its 3 elements, I’ll call the identity by E, the other two
elements by A and B, but G is a group and A and B are 2 elements, so AB must be
in the group, AB is an element in the group, right, because we have a binary
operation on the group, A and B are there, and when you perform the binary
operation on A and B the output is again in G, so AB is an element in the group
which is just E, A, B, so we have 3 possibilities, AB must be E, AB must be A, AB
must be B, so either this or this or this, right, AB is an element of a group which
has 3 elements listed as E, A, B, so it is either E or A or B, so can it be A?

Now I’m going to invoke a very important property that we have in a group, that
we discussed in an earlier video namely, cancellation property, so if AB = A
multiply both sides by A inverse, so we have A inverse AB, A inverse A. Note that
A inverse is also one of the 3 elements but I don’t need to actually consider which
it is, it’s AE, A or B, I don’t care what it is, it is an element in the group so I can
multiply by that, right. which is going to give me by associativity I can do A
inverse A which cancels and I get B = E, right, so the cancellation property directly
gives this, right, because if AB = A, A is nothing but A times E then I cancel A, I
have just spelled it out but this is what we have, but is B = E? Certainly B is not
equal to E because they are distinct elements, so this is not possible, so this is not
possible I’ll say that.

44
So AB cannot be A, similarly AB cannot be B because the cancellation property
says B is equal to BE, I cancel B and I get A = E, right, but A can’t be equal to this,
this can’t happen, this is not the case because the three distinct elements are called
E and A and B, so A can’t be same as E otherwise we would not have given it a
new name so this can’t happen, so AB must be E, but if AB is equal to B, AB = E
rather, AB = E so this means that B is A inverse, remember that if AB = E, B is A
inverse and automatically BA must be E, so A and B commute.

So the point here is that, the important point, just like I said earlier that every
element commutes with the identity, an element and its inverse always commute
with each other, that is the point, so in this case A and B are inverses of each other
and they commute, so AB = BA. If AB = BA, then G must be abelian, right,
because if G remember if G is not abelian there must exist 2 elements in it which
don’t commute with each other, in our case G is a group of 3 elements, E
commutes with A, E commutes with B, only possible violation of abelianness is if
A and B don’t commute with each other but we’ve just concluded that, A and B
commute with each other so G is abelian, so we are done with case 3.

So now case 4, so that means order is 4. So let me now do the following. so I’m
going to write G as, okay, so G has 4 elements right, let’s call them A, B, C, the
non-identity elements are A, B, C, but now I will do the following, suppose G is
not, actually let me take that back, so I don’t want to list enumerate this elements at
this point, so suppose G is not abelian, right, suppose G is not abelian. Then by the
point that I mentioned at the beginning of the proof or beginning of the solution
there exist A, B in G such that AB is not equal to BA. This is the definition of
group G not being abelian, but now, G has only 4 elements, that’s what we are
going to invoke, what would be the elements of G? E is an element of G, and A, B
are second and third elements of G must be, E, A, B must be 3 distinct elements of
G.

Why are they distinct? See ,because A and B do not commute with each other,
neither of A and B can be E, because if A is E, B and E will commute, similarly if
B is E, A and B will commute, because every element commutes with E, and A, B
do not commute with each other they are different from E. Not only that, A and B
are also distinct, right, because if A and B are equal, note A must be different from
B. Why is that? Because if A = B then what is AB? Then AB is just A squared, but
that is also same as BA, so A and B commute, right, but I am assuming that A and
B do not commute, so A and B must be different, so E, A, B must be 3 distinct
elements of G, but G has only 4 elements, so there is room for only one more
element and I claim that must be AB.

45
So now what must be AB? By the previous case, we have said that AB, so what can
be AB? AB has to be E, can it be E? Let’s explore that, can it be A? Can it be B?
Okay, so can AB be E? If AB is E, A and B are inverses of each other, right, if AB
is E they are inverses of each other and hence though they commute, but that’s
clearly wrong because I’m assuming that they do not commute, so AB can’t be E,
AB can’t be A because of the same cancellation property that we discussed in the
previous case, B must be E, but B is not equal to E so that can’t happen.
Similarly in this case A must be E which can’t happen, so AB, so the conclusion is,
so we conclude AB must be the fourth element of G, so G must be E, A, B, AB,
right because we already have 3 elements E, A, B, and AB can’t be any of them,
AB can’t be E, AB can’t be A, AB can’t be B, so AB must be the fourth element,
okay, but now what has to be BA, but what is BA? BA is also in the group, right,
so BA can be E, but if BA is E the same problem happens, B and A are inverses of
each other and hence they commute which is not possible, so BA can’t be E, BA
can’t be A for the same reason as before, and BA can’t be B, so BA must be AB,
right, so again contradicting the hypothesis. So we cannot have, in other words, so
is it clear? So BA must be one of the 4 elements because there is no more room in
the group, group has only 4 elements and we have already identified 4 distinct
elements, but BA cannot be E, because in that case they are inverses of each other
and they commute, BA can’t be A, BA can’t be B, and BA has to be AB in which
case they commute with each other, so it’s a contradiction, so the hypothesis that
we started with that there exist 2 elements AB such that AB is not equal to BA is
wrong, in other words G must be abelian, so case 4 is also done.

So in case 4 we have concluded that G is abelian, so finally case 5, so I want to

show that in the case of order 5 group, it is abelian, so what I will now do is the
following. So I will use the properties that I mean the work that I have already
done. So let’s say E, okay, so again I don’t want to start with this, so suppose G is
not abelian, right, if G is not abelian, then this is exactly as in the case 4, there exist
2 elements A, B in G such that AB is not BA, okay, so now I claim that we must
have that the 5 elements, recall that G has only 5 elements, of G must be E, the
identity element A, B, AB and BA, why is this? Remember that A and B are
elements that don’t commute with each other, so they are different from E, so you
have 3 elements here and just as in the case 4, AB cannot be E because if AB is E,
A and B commute with each other, being the inverses of each other, so it can’t be
E, it can’t be A, it can’t be B, so AB is the fourth element, but now in the case 4
that’s done and BA must be one of them, and it has to be abelian, but now we have
room for one more element, BA cannot be E because B and A are not inverses of
each other, it can’t be A, it can’t be B, and it is not equal to AB also because AB

46
and A and B do not commute, so in other words, this is G, so G is, okay, so this is
our first observation, G must be like this.

We are going to get a contradiction, so I am going to make a series of observations.

So first so I’ll claim that ABA, okay, so what is ABA? Remember that I want to,
okay so maybe I will write it as a claim ABA is B and BAB is A, so this is my
claim, so what is the proof of this? So and this is exactly as before, so I am going
to systematically eliminate all other possibilities, so I’ll do one of them, the other
being exactly similar to this. So what is ABA? Okay, remember ABA is an element
of the group, so it must be one of E, A, B, AB, or BA, so suppose, we have 5
possibilities, right, so ABA is A let’s say, okay, so then just look closely at this
equation, so we can read this in two ways, so we have A times BA is A, and AB
times A, sorry I’m considering the case E, so I am going to consider each
possibility, ABA must be an element of the group, so it is one of this 5 elements, so
suppose it’s E, so A times BA is E, AB times A is E, right, but this is a problem
now, because if A times BA is E, then BA is A inverse and here AB is A inverse,
right. This equation suggests that BA is A inverse, this equation suggests that AB is
A inverse. But remember, when we discussed properties of groups in an earlier
video we have shown that A inverse is unique for any element in a group, the
inverse is unique, but A inverse is, both BA and AB but this means that AB is equal
to BA, which is not the case, right, so ABA can’t be E, so this cannot happen. Can
it be, I’m claiming it’s B, so let’s say, can it be A? So this is not possible, correct, is
it clear? AB and BA can’t both be inverses of A, so ABA is not E, so that we have
concluded.

Can ABA be equal to A? So I have, ABA has 4 possibilities, right, E, 5

possibilities, E, A, B, AB, BA, it has no choice, it has to be one of these 5
elements, I have concluded that it can’t be A, can it be A? It can’t be E, I have
concluded it can’t be E, can it be A? It cannot be, because if I cancel A and
conclude that BA is E, but if BA is E, AB must also be E, because A and B are
inverses of each other and they commute, so it can’t be A. Can it be AB? If it’s AB,
then I cancel A, then I cancel B, I get A = E, this is also not possible, so it can’t be
AB. Can it be BA? It can’t be, again I cancel B and A and I have A = E, right, so it
can’t be B, so ABA must be B, this is the only possibility left, so ABA is B, this is
proved. And exactly the same proof gives me BAB = A, I have to systematically
eliminate all possibilities, it can’t be E, it can’t be B, it can’t be AB, and it can’t be
BA, so let me not prove that, because it’s very similar to this, so I have shown that
ABA = B, BAB = A.

Next I’ll claim that so I think I called it 1, I’ll claim 2, A squared = B squared I

47
claim this. So claim A squared = B squared, why is this true? Remember I have
shown that ABA = B, let me multiply by B on the left, what do I get? So I get
BABA = B squared, I have also proved in the first step BAB = A, now multiply by
A on the left, on the right, so I get BABA = A squared, right, so BABA = B
squared, BABA = A squared, so A squared = B squared.

Now the third part of the proof, I want to show that there is no room for A squared,
so let’s say A squared, what is A squared? okay, so that’s what I want to now do, A
squared if A squared, so again there are 5 possibilities, right, E, A, B, AB, BA, so
let me immediately rule out everything other than E. Suppose A squared is A, right,
this means I cancel A, one copy of A so that means A = E, because A squared is A
star A = A, so that means A = E, so that’s not possible because A is the distinct
element. Can A squared be B, if A squared is B, A squared is also equal to B
squared, right, so B squared is B, because A squared is B squared, and A squared is
B, B squared is B, that means B is E, that’s not possible, so A squared cannot be B,
A squared cannot be B. Can A squared be AB? If A squared is AB I cancel A on
both sides to get A = B which is also not possible, A and B are distinct elements, so
A squared cannot be AB. Can A squared be BA? That would give me, cancelling A,
again A = B which is also not possible, so A squared cannot be A, B, AB, or BA, so
A squared must be A, okay.

Now if you recall the first part what we have shown was recall BAB was A, this
was proved, right, in the earlier part, BAB = A, now let’s multiply by B on the left,
now on the right, if I multiply by B on right I get BABA is AB, right, but BA, sorry
I am multiplying by B, so BABB is AB, but what is BB? So this gives me BA, B
squared is AB, but B squared is also E, if A squared is E, that’s equal to B squared,
so I have shown that A squared = B squared, so BA B squared is AB, but B squared
is E, so BA = AB, but this is a contradiction because I have started with elements
AB which do not commute with each other so AB and BA cannot be equal to each
other, so again we conclude that G is abelian,, G being any group of order 5, okay.

So just to recap what we have done, we wanted to show that any group of order 5
is abelian and we have started considering order 1 which was very easy, order 2
was also very easy, order 3 was also fairly easy, order 4 was not difficult also
because we easily concluded that there is not enough room, order 5 required a little
bit of work but again we conclude that, by systematically using all the properties
that a group satisfies, that G is abelian, so I wanted you to focus and I wanted to do
this in detail because it’s a very good example, it illustrates the key points of a
group, and the point that you must keep in mind and I keep emphasizing in this
course is that we are studying abstract groups, okay, these are groups though, many

48
of the examples of groups that we know are familiar to us integers, real numbers,
rational numbers, functions, rotations, roots of unity, complex numbers, the point is
these are only examples. We want to develop a theory for abstract groups, so we
cannot and we should not use any property that the specific examples have, we
should only use properties that a group by definition has, namely that it has an
inverse, namely that it has identity, namely that binary operation is associative, that
it’s closed, and those are the four important axioms, definition, they are part of the
definition and after that you’ve concluded that there is a unique identity, there is a
unique inverse for every element, using these properties we are able to, and only
using these properties, so if you recall and revise what happened in this video,
nowhere have we used the property that it’s an integer or real number or a function,
everywhere we have only used group-theoretic properties. And we have concluded
that in order to achieve non-abelianess, in other words in order to produce two
distinct elements that do not commute with each other you must go up to 6 order at
least, and in order 6 we do have a non-abelian group, namely S3. So please go
over this carefully and make sure that you understand everything here, and that
will be a good way for you to make sure that you are comfortable with the basic
definition and properties of groups. And I’ll stop this video now, in the next video
we are going to learn about subgroups of a group and study properties of
subgroups, and look at various examples. Thank you.
Online Editing and Post Production
Karthik
Ravichandran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh Kumar
Web Studio Team
Anitha
Bharathi
Catherine
Clifford
Deepthi
Dhivya
Divya
Gayathri
Gokulsekar
Halid
Hemavathy

49
 Jagadeeshwaran
Jayanthi
Kamala
Lakshmipriya
Libin
Madhu
Maria Neeta
Mohana
Muralikrishnan
Nivetha
Parkavi
Poonkuzhale
Poornika
Premkumar
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirley
Sorna
Subash
Suriyaprakash
Vinothini
Executive Producer
Kannan Krishnamurthy
NPTEL CO-ordinators
Prof. Andrew Thangaraj
Prof. PrathapHaridoss
IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resources Development
Government of India
www.nptel.ac.in
Copyrights Reserved

50
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
MODULE – 02
Lecture – 08 - “Subgroups”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

Okay, so in this video I am going to talk about subgroups of a group, we have seen
some examples of these already so I am going to define these and then we will look at
examples and properties. So a subgroup of a group is simply, when I’ll say it words
first, a simply a subset of a group which has the properties of a group which has the
properties of a group.

So let’s fix a group, let’s say G is a group, a subgroup of G, so let’s say subgroup H of
G is a subset of G, so to begin with it’s a subset of G which has the following
properties, which has the following properties. One, H is closed under the operation
of, so this means remember that if A and B are in H this implies AB is in H.
Remember this is really not always true, because if A and B are in H they are in G, so
AB is in G, but here we want it to be in H again, so it’s a condition on H, so it must
be closed under binary operation of G, the identity element must be in H, and, three,
if A is in H then A inverse should be in H, okay, so this is the subgroup, so that’s all,
so that is the definition of a subgroup.

The third condition here is saying that if an element is in H it’s inverse is also in H,
again remember that inverse is in G but it need not be in H, we want it to be in H, so I
will only remark here, an easy remark, if H is a subgroup of G, then H is also a group
under the same operation as, so if you focus your attention only on H it’s actually a
group because it has a
binary operation, same as the one of G, it has identity element, it has inverses, but the
group has another property, right which is associativity, but remember that
associativity comes for free in this case, because associativity holds in G, so if you
get 3 elements A, B, C in G you can group them in any of the two ways to get the
same answer, hence if 2 element, 3 elements are given in H, associativity holds
automatically for them because they are elements of G, so it’s also a group, we do not
need to check for associativity separately, so some immediate examples and non-
examples really.

Z is a sub group of Q under addition, so this is obvious, right, because under addition
Q is a group to begin with, Z is closed under addition, it has 0, and it has inverses,
okay.

So similarly Q is a subgroup of R or C again under addition. Q star which is

remember nonzero rationals is a subgroup of R star or C star, these are nonzero reals,

51
these are nonzero complexes, this is under multiplication. I am not checking this in
detail but it’s very clear, right, Q star is subgroup of R star because it is closed under
multiplication, it has identity, it has inverses.

What about Z star? Is not a subgroup, for example of Q star, nonzero integers do not
form a subgroup under of Q star under multiplication, because it is closed under
multiplication, it has identity but it doesn’t have inverses. See, note that the reason is
same as the reason that we saw in an earlier video that Z star is not a group, so it can’t
be a subgroup of Q star.

Okay, so some more interesting examples, let’s take the subset of Z consisting of
even integers, so in other words H is 2N as N varies over Z, right, even integers are
divisible by 2, so they are always multiples of 2 so you can write them as 2 times N,
as N varies over Z. Is this a subgroup? Is this a subgroup of Z? Let’s check one by
one, what are the properties? Is even integers closed under the binary operation
which is of course, I haven’t emphasized here because it’s clear generally I am
considering Z under addition, so I take Z under addition and take H to be the even
integers.
Are they closed under the addition of integers? Yes, if you sum two even integers it is
even integer, right, is the identity element in the set H, yes, because 0 which is the
identity element is an even integer. Similarly if N is an even integer it’s inverse, what
is the inverse under addition? It’s minus, okay so minus of an even integer, it is yes, it
is a subgroup and the reasons I’ll not write this, but I’ll just say that 0 is an even
integer, sum of even integers is an even integer, negative of an even integer is an even
integer, so even integers form a subgroup.

What about odd integers? So take the subset of odd integers, is it a subgroup,
certainly it’s not a subgroup because it’s neither closed nor does it have an identity
element, because you can have sum of two odd integers, and you get an even integer
3 + 3 is 6, 3 is odd but the sum of 3 and 3 is 6. Similarly 0 is not there so because 0
for example is not odd, so odd integers do not form a subgroup but even integers do.

What about this set 3N, N in Z? Earlier we took 2N which is the even integers, right,
that’s a subgroup we saw, what about odd, multiples of 3, so these are all multiples of
3, so this set is actually -6, -3, 0, 3, 6, 9, 12, and so on. That is this set, this is also a
group, this is a subgroup of Z, because 0 is there, sum of any two things here will be
again a multiple of 3, if you take 2 multiples of 3 and sum them, for example 3 + 6 is
9, which is also a multiple of 3 but more generally if you do 3N + 3N it will be 3
times N + M, so it is closed under addition, and negative of multiple of 3 is also a
multiple of 3, so it is a subgroup of Z, more generally AZ which is by definition all
multiples of A is a subgroup of Z for any A in Z, okay, so maybe I’ll write that as a
theorem or, let me wait for the theorem but why is the previous statement true? AZ is
a subgroup for any A in Z, it’s exactly the same reason as it is for 2Z and 3Z,
remember this is nothing but 3Z, and even integers are nothing but 2Z, so all
multiples of 2, so this 2Z is just my notation, more generally AZ is the collection of

52
all multiples of A, it’s a subgroup of Z for every small A, because if you take 0, N as
0, 0 is inside that so let me just check by one by one, so AZ is closed under addition,
why? Because if you take AN + AM this is a multiple of A, this is a multiple of A,
you are adding them you get A times N + M which is also multiple of A, so that is
okay, AZ contains 0 certainly, because 0 is A times 0, remember AZ is all multiples of
A, so you can take AN for every N in Z so in particular when you take N = 0 you get
0. And similarly AZ is closed, contains inverses that is because inverse of AN is A
times – N, so that’s also inside AZ, so AZ is, so these 3 points imply that AZ is a
subgroup of Z. Now the theorem that I want to write and prove is that every subgroup
of Z is of the form AZ for some positive integer, okay, so for some integer not
positive, but some non-negative integer A, okay, so this is the theorem so I am, earlier
before the theorem I showed that A times Z is a subgroup for every small A, but now
I making the converse statement, I’m saying that every subgroup of Z is of the form
AZ, this specific kind of a subgroup, this requires a proof, right, these look like
specific kinds of subgroups, but why should every subgroup of Z be of that form, so
let’s prove this.

So how do I prove this statement? I am going to prove that any subgroup you give me
of Z must equal AZ, where A is some nonnegative integer, so let me prove it by first
starting with an arbitrary subgroup of Z, let’s say H is a subgroup of Z. So we are
going to first assume that, consider the case that H is 0, that’s certainly a possibility
right, 0 by itself is a subgroup of Z, if H is just 0, then is it of the form AZ? Yes, it is
of the form 0, Z because this is 0 times N, N in Z that means this is just 0, so
remember I’m allowed to take any nonnegative integer A, I’m going to take 0 in this
case, if H consists only of the zero element it is 0Z which is of the required form, so
that case is done.

So now, so suppose that H is not just the zero element, so H contains an integer N,
right, which is different from 0 because H is a subgroup of Z, it is not just the single
element 0 that means it contains a nonzero integer, yes, so it contains a nonzero
integer. I claim that in fact H contains a positive integer, why is this? Why? Okay, so
I first said that H contains an integer N nonzero, so let N nonzero be an element of H,
I know that there is a nonzero element because remember I have assumed that H is
not zero, H = 0 case I have already settled, so H is not equal to the single element 0,
so it contains an element N of which is different from 0. If N is positive we are done,
we are done in the sense that we are done with this statement, I’m trying to prove that
it contains a positive integer, so N itself is positive we are done, if N is less than 0
remember N is different from 0 so N is either strictly more than 0 or strictly less than
0, if N is less than 0, – N is positive, right, if N is less than 0, – N is greater than 0,
but I claim that since N belongs to H, – N also belongs to H, why is this? This is
because H is a subgroup, this is where we have used the fact that H is a subgroup,
remember if you recall the definition of a subgroup, if A belongs to H, A inverse
belongs to H, the inverse of A belongs to H. In my example, I’m working with Z
under addition so inverses are negatives, if N is H, so N is an H, -N is also in H, so if

53
N is negative, – N is positive. So I have justified this statement here that H contains a
positive integer N.

Now define or let A be the smallest, this is a very important argument, you should
pay close attention to this, this comes up a lot in algebra, so I have first said H
contains a positive integer, now I’m going to take this smallest positive integer
contain in H, remember that H contains perhaps lots of positive integers, but there
will always be a smallest positive integer because any set of positive integers has a
smallest element, so if you take the set of positive integers in H it will contain a
smallest element, so I take A to be the smallest positive integer in H, so how do I get
A? So I’ll start with 1, is 1 in H, if not I’ll go to 2, is 2 in H, if not I’ll go to 3, if 3 is
not in H I’ll check with 4, otherwise I’ll check with 5, because H contains a positive
integer at some point we will reach a number which is in H, the first time we reach
that is the smallest positive integer that’s in H, so I’ll call that A.

Then we claim that H must contain only multiples of H, A is my claim which also
proves the theorem, so why is H = AZ? So now let, let’s say B is a positive integer in
H, so let’s say, let me write it like this, let B in H and assume, okay, so let’s say B is
in H and assume B is positive, so I am going to consider this case. By the choice of
A, what is choice of A? A is the smallest positive integer contained in H, B is some
other positive integer contained in H, so by choice of A, we have B greater or equal to
A, B could be A of course, but it can’t be smaller than A.

So now we divide B by A, so you all know division, what does it mean? So if I divide
B by A, I can write it like this, B = some A times some P + Q, so if I divide an integer
by another integer, I’ll have some reminder, what is the properties of this P and Q? P
is in some, P is some Z, element of Z and Q the important property is Q is strictly
between, sorry, Q is nonnegative but it is strictly less than A, see remainder is always
less than A, correct, because if Q is more than or equal to A, I can further divide,
okay, so I can keep dividing until the remainder is strictly less than A, so I have this.

Now this equation will translate to B – AP = Q, now let’s observe this closely. B is in
H, that is by hypothesis, A is of course in H because A was chosen to be, remember
that A was the smallest positive integer contained in H, so we claim A is in H, B is in
H again using the properties of a subgroup, so note that A is in H, so –AP is in H,
because H is a subgroup, – AP, what is –AP? So actually if you want to do step by
step, if A is in H, why is – AP in H? –AP remember is P times –A, I can write it like
this, that means it’s – A + -A + -A, P times. A is in H, so –A is in H, that is the
property of a subgroup, if –A is in H, P times you add –A to itself that’s also in H, so
that’s the proof for this. Similarly B is in H, that is also given, hence B –AP in H. So
A is in H, -AP is in H, and B is in H so that B + -AP is also in H, so B – AP is in H,
then Q is in H, right because B –AP =Q, so Q is in H. But now let’s see something
interesting happens, Q is in H but Q is strictly less than A, Q is strictly less than A,
can Q be positive now? It cannot be, Q can’t be positive, because Q is strictly less
than A and A is the smallest positive integer in H, and Q remember is the conclusion

54
at this point, is Q is in H, so Q can’t be, if Q is in H and it is positive, and it is less
than A, that violates the choice of A, A was chosen to be the smallest positive integer,
so Q must be 0. Then if Q is 0 the remainder is 0, if Q is 0, this is 0, so B = AP, so if
B is AP this means that B is inside AZ, see I have started with an arbitrary element of
H which is positive and concluded that it is a multiple of AP, a multiple of A, I don’t
care what P is, it’s just a multiple of A, that’s what I am interested in, so every
positive number in H is a multiple of A.

What about negative integer? So if, let say B is a negative integer, if B is in H and B
is less than 0, consider –B which will be positive, -B is in H because B is in H, and H
is a subgroup, -B is in H and by what we have already shown which is that every
positive integer is a multiple of A, –B is a multiple of A for some P, but then B is A
times –P, so B is also a multiple of, in other words, so we have showed that, shown
that every element of H is a multiple of A, hence H is contained in AZ, remember AZ
is the set of all multiples of A, every element of H is a multiple of A, so H is in
contained in AZ, but clearly, because, two reasons, H is a subgroup and A is in H, A is
in H because remember A is the smallest positive integer that’s in H, so A is in H, H is
a subgroup so twice A is in H, thrice A is in H, and hundred times A is in H, so all
multiples of, positive multiples of A is in H, because A is in H, -A is in H and all
positive multiples of –A are in H, so entire AZ is contained in H, so AZ = H,
remember that’s what we are trying to prove, so this completes the proof. This
completes the proof of the theorem that we have every subgroup, so I think the
theorem is here, what is the theorem, so let’s see, every subgroup of Z is of the form
AZ for some nonnegative integer AZ, for some nonnegative integer A in Z.

So this is very strong property of subgroups of Z, so I want to just make one more
remark as I said this is an important proof, okay so in this proof as I said the proof is
very important to understand, and just to make emphasize my point what we have
really done is you take the integers, 0 is here, and I’m looking at a subgroup of Z, I
have taken the smallest positive integer, so A is the smallest positive integer in H, so
that means there is, between 0 and A there is nothing in H, so as you go from 0 to the
right first time you hit H is at A, okay, so then next time you hit H is 2A, so there is
nothing in between A and 2A that can be in H, that is because if there is something in
between by subtracting A you land here, but then we know that there is nothing
between 0 and A in H.

Similarly at the, next one is 3A, next one is 4A, so H must be only multiples of, so
similarly between –A and 0 there can’t be anything in H because if there was
something here in H, it’s negative will come here and that will violate the property
that A is the smallest positive integer, so in some sense H must be just this made up of
this multiples of A, okay, and nothing in between these multiples can be in H, so this
is exactly what we have done. So subgroups of Z are particularly simple in this way.

One more remark. Every group has two obvious subgroups, right, no matter what the
group is, it has two obvious subgroups. You can take E is a subgroup of G, definitely

55
it’s a subgroup of G because it is closed under the binary operation, there is just one
element so if you apply E squared E cube it’s all E, so it is closed under the binary
operation, the identity is there inverses are there, so it is subgroup. Similarly, this is
called the trivial subgroup, okay, so it is trivial but it has just E. G is also a subgroup,
G is a subgroup of G also, because it is certainly a group, so and it’s closed under the
binary operation by definition it has inverses, it has identity, so it is the full group, so
these are not interesting, so typically we are interested in subgroups which are neither
the trivial subgroup or the full subgroup.
Online Editing and Post Production
Karthik
Ravichandran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh Kumar
Web Studio Team
Anitha
Bharathi
Catherine
Clifford
Deepthi
Dhivya
Divya
Gayathri
Gokulsekar
Halid
Hemavathy
Jagadeeshwaran
Jayanthi

56
 Kamala
Lakshmipriya
Libin
Madhu
Maria Neeta
Mohana
Muralikrishnan
Nivetha
Parkavi
Poonkuzhale
Poornika
Premkumar
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirley
Sorna
Subash
Suriyaprakash
Vinothini
Executive Producer
Kannan Krishnamurthy
NPTEL CO-ordinators
Prof. Andrew Thangaraj
Prof. Prathap Haridoss

57
 IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resources Development
Government of India
HYPERLINK "http://www.nptel.ac.in" www.nptel.ac.in
Copyrights Reserved

58
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
MODULE – 02
Lecture – 09 - “Types of Groups”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

Okay, so another important notion is that of a, so let G be a group and let A be in G, let’s take an
element of it, so I want to define the subgroup generated by A is the subgroup, so the subgroup
generated by A so that is the term I’m defining here is this set, so this if you split out it looks like
this, it has E, it has A, it has A squared, it has A cubed, it has this, it has also A inverse, it has A – 1
squared which is under notation A – 2, A -3 like this, okay, so this is denoted by let’s say this
symbol, the subgroup generated by, definition of this is this, so the subgroup generated by A is
denoted by A within these brackets.

First of all, why, this is certainly a subgroup. This is a subgroup that I am trying to define a
subgroup generated by A, why is this a subgroup? So it’s very clear right, because identity is there,
if you multiply two things here it is A power N times A power M so it is, as I commented in earlier
video when I was looking at an example, you can take usual exponential rules, so if you multiply
two powers of A, you get another power of A, and inverses are also there because A inverse is there,
A squared inverse is A -2 and so on, so this is subgroup.

Not only that. If a subgroup H of G contains A then A power M is in H for every integer, right,
because if H contains A and H is a subgroup, the properties of a subgroup say that A inverse is in H,
A squared is in H, E is in H, so A power 0 is in H, A power -2 is in H and so on, so all powers of A
are in H, in other words H contains the subgroup generated by, all this is supposed to imply, and all
this implies that the subgroup generated by A is the smallest subgroup of G containing A. Nothing
smaller than this can contain A, because once it contains A, the group properties guarantee that, if a
subgroup contains A it must contain all of these elements, okay, so it’s important to, this is an
important notion, so subgroup generated by an element.

Okay, so now a very important definition here about groups. A group G is called cyclic, it’s a cyclic
group if there exists an element A in G such that the subgroup generated by A is G, okay, so in other
words, G must be all powers of A, so if G consists only of powers of a specific element then we say
that G is cyclic, okay, so this is an important class of groups. For example, Z under addition is
cyclic. It is cyclic because, what is a subgroup generated by, by what, so namely this, what is this?
This is equal to Z, because you take 0, you take all multiples of 1, so you take 1, 2, 3 times 1, 4
times 1, you take 0, you take negative of 1, you take twice the negative of 1, thrice the negative of
1, four times the negative of 1, this is equal to the subgroup generated by 1, but of course this is Z
also, so Z is cyclic and it is generated by 1, so if G is a cyclic group there is an element A such that
the subgroup generated by A is G, so in this case we say that G is generated by A or A is a generator
of G.

In our example, 1 is a generator of G. In fact, -1 is a generator of Z also, 1 is a generator of Z as I

explained here, but -1 is also generator of Z, because -1 multiples also cover all of Z. Is 2 a
generator of Z? Is 2 a generator of Z? To answer this question consider the subgroup generated by 2,
this is all multiples of 2, right, in the general notation I called it all powers of 2, but here the
addition is the operation so it’s all multiples of 2. And we have a name for this, it is 2Z, but this is
not equal to Z, so 2 is not a generator of Z. In fact no number other than 1 and -1 is a generator of Z,
okay.

59
So now some more notation here, so let us now go back to a general situation. Let G be a group and
let A be an element of it. I want to define the order of A is the order of the subgroup generated by
this, if this is finite, okay, order of an element A is the order of the subgroup generated by that
element, if the subgroup is finite, otherwise if the subgroup is infinite we say that, we say that, the
order of A is infinite, order of A is denoted by, we denote this by order ord (A), so order of A recall
again is it is, it is obtained by looking at the subgroup generated by A, if it is a finite subgroup, if
it’s a finite group we say the number of elements of that is order of A, otherwise if the subgroup
generated by is not finite, then order of A is infinite.

Okay, so as an example G is any group, order of the identity element is always 1, that’s clear right
because what is the subgroup generated by E, is just E, only things it contains are powers of E
which is just E, so order of E is, so order of the identity element is always 1. On the other hand if
you take the integers, what is order of A? If A is not 0 this is infinity if A is not 0, because for any
nonzero element, so this is not, if A is nonzero, if A is 0 order is 1, if A is nonzero order is infinity
because the subgroup generated by that is going to be infinite, because it is AZ it has infinitely
many elements.

So now I want to do this with important example, and you are required to remember from an earlier
video what S3 is so remember S3 is, recall S3 was the group, bijections, group of bijections of a 3
element set which we denoted by 1, 2, 3, and in the video when I discussed this in detail I used the
notation F1, F2, F3, F4, F5, and F6 to denote all the bijections, so I would like you to compute
orders of elements of, okay, and answer the question: is S3 cyclic? Okay, so I won’t do that details
and I will let you do the work, you can stop the video here and do the work and, no, no I don’t mean
you can pause the video here and do the calculations but I’ll give you the answer in case for you to
check, so if you remember the notation of this F1 to F3 from F1 to F6 from the one of the earliest
videos, so you can remember that F1 was the identity bijection, so order of F1 is F1 is 1. Order of
F2, and order of F3, and order of F4 will all be 2, okay.

So for example if you do the calculation you will see that the subgroup generated by F2 is just F1
and F2, because F2 squared is just F1. F2, F3, F4 were permuting 2 of the 3 numbers, and third one
is kept fixed, so F2 fixed 3 and sent 1 to 2, 2 to 1, so F2 squared will be identity, so the subgroup
generated by F2 is just {F1, F2}, so order of F2 is 2, so this is similar to, the same is true for F3 and
F4.

On the other hand if you look at F5 and F6, the subgroup generated by F5 is actually {F1, F5 and
F6}, I also asked you in an earlier video to completely describe the multiplication table of this, and
if you did that it would be clear to you that subgroup generated by this is this, so ord and which is
also same as subgroup generated by F6. so order F5 and ord F6, so you have 3 elements, you have 1
element of order 1, 3 elements of order 2, and 2 elements of order 2, so S3 has one element of order
1, always every group has one element of order 1, 3 elements of order 2, and 2 elements of order 3.

Now the question that I asked earlier: is S3 cyclic? Okay, so remember from an earlier slide, what is
a cyclic group? So somewhere I missed this, yeah, so a subgroup, a group is cyclic if there exists an
element A in G such that A is the generator of G. So now is S3 cyclic? We have 3 possible elements,
do they generate S3? Does F1 generate S3? Certainly not because F1 is identity, does F2 generate
S3? No, because F2 the subgroup generated by F2 is just {F1, F2}. Does F3 generate S3? No, again
order is 2, so similarly F4 does not generate S3, but neither do F5 and F6, so because the subgroup
generated by F5 is simply 3 elements, not all of S3, similarly subgroup generated by F6 is 3
elements, so S3 is not cyclic, that is a point.

60
So as this example shows I’ll give this as an exercise and I will let you work this out in detail, it’s a
good exercise to understand the concepts. A group, let’s say a finite group, let G be a finite group,
let’s say, of order N. So it has N elements, then G is cyclic if only if G contains an element of order
N, because remember G is cyclic means G must be equal to the, equal to a subgroup generated by
some element, G has N elements, so if G is cyclic G must contain a sub element, subgroup
generated by which has N elements, so that element is order N. If G contains an element of order N
then the subgroup generated by that element will be of order N, but G already has N elements so the
subgroup also has N elements, so G must be equal to that, okay so the proof I just said orally you
should combine all the details I said and conclude that, a finite group is cyclic if it is only if it has
an element whose order is equal to the order of the group.

Okay, so also another exercise for you, and this one I will quickly workout. Suppose that G contains
no subgroups different from {E} and G, so note that every group as I remarked earlier, every group
contains 2 obvious subgroups, namely the trivial subgroup and the full subgroup, suppose that some
group G does not contain any other subgroup, okay, then G is cyclic.

Why is this? So I’ll write the solution for this, okay, so suppose we take an element, okay, so
actually if G is actually all of E, G is a single element group, then it is cyclic, right because this is
certainly cyclic. So assume that G is different from {E}, so that means G contains some element
different from E, so let say A is in G, A different from E. Now consider the subgroup generated by A
which is, which we denote by this symbol, so it cannot be equal to this, right, because A is not equal
to E and the subgroup generated by A certainly contains A so it’s not equal to {E}.

Now what is the hypothesis on the group G? It says that G contains no subgroups different from
{E} and G, right, this is a subgroup, A the subgroup generated by A is a subgroup, it is different
from {E}, so it must be equal to G, right, because G contains subgroups, no subgroups other than
the trivial subgroup and the full group and we have constructed the subgroup which is different
from trivial subgroup, it must equal all of E, so G is cyclic by definition, right, as soon as you have
an element whose subgroup generated by which is equal to G, G is a cyclic, so that completes the
solution.

Okay, so I will now give you some important subgroups for any group, so now let G be any group,
okay, so I am going to define a few subgroups which are very important and we will study them
later. The first one is the center of G denoted Z(G), it is all elements, let’s say g in G which have the
property that Ag = gA for every, okay, so you need to stare at this for a minute to understand this
carefully. Center of G denoted by this symbol is the collection of elements of the group which
commute with everything else, so Ag must be equal to gA for every small A in G, okay, so the
proposition is that Z(G) is a subgroup, so let me prove this.

The previous slide contains the definition of Z(G), what is a subgroup? We must show that it is
closed under the operation of G, so let’s say G1 and G2 are in Z(G) to prove, we want to prove
G1G2 is in Z(G), right, we want to prove this, what does this mean? In order to be in Z(G) it must
commute with an arbitrary given element of G, so let’s take an arbitrary element of G and see what
happens to G1G2 times A, this by associativity of the group is G1G2 of A, right I can put the
bracket here, but G2 remember is in the center, so G2 commutes with A, so this is G1 of times AG2,
but again applying the associativity this is G1A times G2.

Now I’ll continue here. G1 is in the center, let’s use that now, so it is AG1 G2, again using
associativity it is AG1 G2, so now we are done because G1 G2 times A is A times G1 G2, so that
tells me that G1G2 is in Z(G), so Z(G) is closed, is the identity element in Z(G)? It is certainly there
because E times A is A times E = A for A and G, so certainly the identity element always commutes
with everything so that is okay, and finally if G is in Z(G), is G inverse in Z(G), let’s check, so we

61
want to check G inverse times A must be equal to A times G inverse, so what is this? So this is, if
you do, so I claim that this is equal to A inverse G inverse, so if you recall I gave an exercise or I
mentioned this in an earlier video, in any group AB inverse is B inverse A inverse, so this is, I’m
applying that here, so basically applying this, G inverse A is the inverse of, I have to interchange the
letters here, so it’s A times, inverse of A times G the whole inverse, okay, so this is correct, right.

But G is a element in the center, so G commutes with everything, so this is G times A inverse,
nothing I only interchanged within the bracket I interchanged these two, but now again applying
this, this is GA inverse whole inverse is A times G inverse, okay, so G inverse is in Z(G), so hence
Z(G) is a subgroup, this completes the proof. So the center is a subgroup consisting of all elements
that commute with everything. So as an easy exercise for you if G is abelian then what is the center?
Center is the set of elements that commute with everything, so certainly it must be everything, okay,
so this is not surprising.

One final thing I’ll define in this video and we will stop after that. So let G be a group and let A be
an arbitrary element, the centralizer, so this is the definition, centralizer of A, denoted C(A), is C(A)
is all elements of the group which commute with A, okay, so earlier when I defined the center, it is
elements of G that commute with everything, AG = GA for every A in G, now I don’t care about
everything, I fixed A and I only define it to be all things that commute with that specific element, so
as before and this I’ll leave for you as an exercise and prove it maybe in the next video, C(A) is a
subgroup of G, the center is always contained in C(A) for all A in G, okay, and also if G is abelian it
is clear that C(A) is equal to G for all A in G, okay, so these are three exercises for you and I’ll
maybe comment on some of these in the next video, so I’ll stop now, in this video, we’ve looked at
definition of a subgroup, we’ve looked at various examples of subgroups, we have shown that all
subgroups of Z are obtained as multiples of a fixed integer, we’ve looked at subgroups generated by
an element and defined the order of element, cyclic groups and finally we have defined center of a
group and centralizer of an element of a group.
Online Editing and Post Production

Karthik

Ravichandran

Mohanarangan

Sribalaji

Komathi

Vignesh

Mahesh Kumar

Web Studio Team

Anitha

Bharathi

Catherine

Clifford

62
 Deepthi

Dhivya

Divya

Gayathri

Gokulsekar

Halid

Hemavathy

Jagadeeshwaran

Jayanthi

Kamala

Lakshmipriya

Libin

Madhu

Maria Neeta

Mohana

Muralikrishnan

Nivetha

Parkavi

Poonkuzhale

Poornika

Premkumar

Ragavi

Raja

Renuka

Saravanan

Sathya

Shirley

63
 Sorna

Subash

Suriyaprakash

Vinothini

Executive Producer

Kannan Krishnamurthy

NPTEL CO-ordinators

Prof. Andrew Thangaraj

Prof. Prathap Haridoss

IIT Madras Production

Funded by

Department of Higher Education

Ministry of Human Resources Development

Government of India

HYPERLINK "http://www.nptel.ac.in" www.nptel.ac.in

Copyrights Reserved

64
 NPTEL

NPTEL ONLINE COURSE

Introduction to Abstract

Group Theory

Module 02

Lecture 10- “Group homomorphisms and examples”

PROF. KRISHNA HANUMANTHU

CHENNAI MATHEMATICAL INSTITUTE

Okay, so in this video we are going to learn about homomorphism

of groups and talk about when two groups are same. Ok, so let’s
start with this definition of a group homomorphism. So,

(Refer Slide Time: 00:34)

let G and G prime be two groups. So, I am going to write the

definition and then we will discuss some examples. So, let G and
G prime be two groups, so this is the definition. A homomorphism
of groups, so it is a homomorphism of groups, is a function, so
usually we denote this by phi (φ) ok, this is read as phi. Φ is a
function from G to G prime it’s a set-theoretic function first of all.

It’s a function from G to G prime such that it satisfies the

following condition. Φ of (ab) is equal to φ of a times φ of b, φ
(ab) = φ (a) φ (b). That’s all. So, we want φ of a, b equals φ of a
times φ of b. Okay, so let us just recall our convention here when I
write ab remember this is for all what are a & b you have to ask is
true for all a & b in G, so a & b are elements of the group G, the

65
domain group and in the bracket here I have ab which means a
times b in the group G. I apply the binary operation of the group G
to a & b.

Φ of a and φ of b are elements of the group G prime and here I

write φ of a times φ of b and that really means that I am applying
the group operation of G prime. So, the left hand side group
operation of G is being used and right hand side group operation
of G prime is used. Okay, so let’s keep this is mind. So, it is not
any set function but a set function which satisfies this. A good way
to remember this is to say this is a set function which satisfies or
which respects the group operations of G and G prime.

Whether you multiply first, so this is the way you should think of
this whether you multiply first in the group G and then apply φ or
you apply φ first and then multiply in the group G prime, you get
the same answer. This is what a group homomorphism is. Okay, so
whether you multiply first and then apply G or apply G multiply
and then apply φ it is same as apply φ and then multiply.

So, multiply and apply φ. Remember when I use the word

multiply I really use it to stand for the binary operation of the
group G, it’s not multiplication of ordinary integers or rational
numbers. Right, because G is an arbitrary abstract group. So,
multiply is just a convenient word to indicate the binary operation.
So, multiply and apply φ that is what we have done in the left
hand side is same as apply φ and multiply, which is what we have
done on the right hand side. We applied φ first and then
multiplied, so we started with a & b we first multiplied them and
then applied φ to get φ of ab. On the right hand side we have apply
φ first to a and b separately and then multiply. So, a group
homomorphism is one were these two operations are going to give
you the same answer. Okay let us start with some after the
definition let us look at some basic examples and before we
continue study the properties of

66
(Refer Slide Time: 04:53)

group homomorphisms. Let us look at some examples. If you look

at, okay let us look at this example. Let us look at our group that
we have studied in detail, the group of integers again when I write
Z and do not say what the operation is, it is understood that the
operation is addition. So, for example let’s say I take an integer a
and multiply by so you fix an integer n and I define φ of a to be n
times a.

So, this is nothing but a + a + a…, n times. Is this a group

homomorphism? Is this a group homomorphism? Let’s check. So,
remember what is that we have to check? We have to check that if
you apply the binary operation first and then if you perform the
binary operation and then apply φ it should be same applying the
binary, apply the function and then perform the binary operation.

So in this example it amounts to checking, we want to check a and

b are two integers you want to check that after adding them and
then applying φ or first applying φ and then add. This should be
true for all integers a & b. Right, but in order to check this let’s
individually calculate them. What is φ of a + b by definition? Φ of
a is n times a. So, φ of a + b simply n times a + b. φ of a is na and
φ of, sorry this is actually b okay, φ of a + b should be φ of a + φ
of b and φ of b is nb. so, n times a + b is equal to na + nb, yes. So,
is this a group homomorphism? Yes.

Okay, so this is a group homomorphism. So, this is a good

example of a group homomorphism, multiplying by a fixed integer
n. Okay, let us look at another example. So, let us say I consider φ
so let maybe I keep using the word letter φ for all these
homomorphisms. Let us say I have {1, -1}. I take this to be G
prime. We have not seen this explicitly before but is this a group?

67
Is G prime a group? It is, under our usual multiplication. Right
because 1 is the identity element, 1 is the identity element right
and -1 is the second element -1 squared is -1 squared is 1, -1 times
1 is -1 and so on. So, this is really G prime is actually a subgroup
if you think about it, of let’s say rational numbers under
multiplication. So, I will make that note here and it is for you to
check.

(Refer Slide Time: 08:40)

Note that G prime is a subgroup, subgroup of Q*. If you recall, Q*

is the group of non-zero rational numbers under multiplication.
So, this is G prime is just a subgroup of it. So, I need to identify G
prime as a group right? In order to talk of group homomorphisms
so we need two groups. We know that Z is a group and now I am
considering the group {1,-1}. That is the group. So, what is the
map φ?

So, let us define the map φ, so define φ of A to be so we have two

cases. So, let’s define it to be 1 if A is even, so if A is even I define
φ of A to be 1 and let us say if A is odd, I define it to be -1. So, φ
of A is simply going to keep track of whether the number A is odd
or even. If it is even we send it to 1, if it is odd we send it to -1.
So, it is a function from the integers to the set {1,-1}. Is this a
group homomorphism? Is φ a group homomorphism?

So, again remember one has to check that φ of a+b should be

equal to φ of a + .Okay so now this is a good example because it
will make you deal with the notation that we have been using so
far. If you go back and see the definition of a group
homomorphism I should say φ of AB equals φ of a times φ of b
but remember I have clarified at that point that the multiplication
here is whatever is a binary operation in the group that we are

68
dealing with. So, in this example we have Z on the left hand side
which is a group under addition and G prime which is {1, -1}
which is a group under multiplication.

So, in the definition we have to check that φ of ab is φ of a times φ

of b. Right, this is what we have to check, but in this specific
example, the left hand side is actually φ of a+b because the binary
operation here is a+b the addition should be equal to and what is
the binary operation on the right hand side? It is just the
multiplication, so it is φ of A, times φ of B. is this true for all A, B
in Z. Let’s check this.

Note that the image of φ is either 1 or -1 depending on whether A

is even or odd. So, let us take two even numbers. So, let us say are
both even. If they are both even, φ of A equal φ of B equals 1
because if both are even they both go to 1 and note that then a+b is
also even. So, φ of a+b is also 1. If two even numbers are added to
each other may also have an even number. We again get even
number.

So, φ of a+b is 1 and remember then this condition is satisfied. So,

φ of a+b is φ of a times φ of b because this is 1 and this is 1 times
1. So this is okay. On the other hand if A & B are both odd, then
what do we have then φ of A and φ of B are both -1 but if A & B
are both odd, what is A+B? Sum of two odd numbers is even so
A+B is even then what is φ of A+B? That is 1. Again this
condition is satisfied right because φ of A+B is 1, φ of A, times φ
of B is -1 times -1. So, this is again okay. So, if both are odd also
φ of A+B equals φ of A times φ of B. right. Now let us check the
other case.

(Refer Slide Time: 13:32)

69
So, if A is even and B is odd let us say in this case what happens?
So, we have φ of A is 1, φ of B is -1 because A is even and B is
odd, A+B is odd. Right, A is multiple of 2, B is not a multiple of 2
and when you add you will not get multiple of 2. So, A+B is odd,
so φ of A+B is -1 again remember the function is odd numbers go
to -1 and even numbers go to1. So, in this case again we have φ of
A+B is φ of A times φ of B because this is -1 and this is 1 and this
is -1. So, this is okay. So, similarly if A is odd B is even you can
check that the same condition holds. So, φ is a group
homomorphism.

Okay, so the function which sends an integer to 1 or -1 depending

on whether it is even or odd is a group homomorphism. Okay, so
the next obvious example would be to consider, so this is probably
3, is the function, the previous function modified slightly. So, let
us keep the same groups Z and {1,-1} but now send A to -1 if A is
even, 1 if A is odd. Okay, so I have interchanged the previous
function.

So, here I sent even numbers to 1, odd numbers to -1. Now I have
sent I am sending even numbers to -1, odd numbers to 1. Is φ a
group homomorphism? Is φ a group homomorphism? I will leave
this as an exercise for you. It is a very easy exercise you can check
that it is not a group homomorphism. You can produce two
numbers which violate the definition of a group homomorphism.
Okay, so this is an example of a function which is not a group
homomorphism. So, let me not do anything about this solution
because I would like you to try to do this on your own.

(Refer Slide Time: 16:10)

One more example. Let’s recall Gl n(R) so, I do not know if I used
the same notation before. Let me use G this notation here Gl n (R).

70
Maybe I used some other notation in the past but the set Gl n (R )
remember stands for all invertible n by n real matrices, under
multiplication. Okay, so we take all invertible matrices which have
size n by n and which have real entries and I make it a group by
multiplication because now every matrix in this set has an inverse
because it is invertible. Product of two invertible matrices is
invertible and inverses exist by definition and so on…
multiplication of matrices is associative. Identity is there, so it is a
group. This is the group under multiplication.

Now we all know what determinants of functions are, so I

consider the determinant function. So, let us take φ from Gl n (R) to
non zero real numbers. So, I take a matrix A and I define φ of A to
be determinant of A. So, just simply written det (A). So, det (A) is
the function which sends A to determinant of A.

This is the function. So, is φ a group homomorphism? First of all

before you even attempt to answer this question, you have to ask
yourself what are the two groups involved. I just recall for you
what is Gln(R). Gln(R ) is the group of all invertible n by n real
matrices under multiplication but what is R*? R* is all nonzero
real numbers, also under multiplication right. So, this is also a
group that we have seen before.

If you take nonzero real numbers and consider the binary

operation given by multiplication it’s a group. So, now I can ask:
is this is a group of homomorphism? It is the function from one
group to another group and the answer is yes and it follows from
the properties if you have seen matrices before properties of the
determinant. What do we have to check? We have to check again
remember the definition of a group homomorphism is φ of ab
should be same as φ of a times φ of a.

I am going to use letters capital A, capital B because those that is

more standard to denote matrices by capital A, capital B. Again I

71
recall for you the definition of group homomorphism is φ of ab
must equal φ of a times φ of b. So, is this true? Yes because what
is φ of AB this is determinant of AB and what is φ of A times φ of
B? This is determinant of A times determinant of B and if you
have studied matrix theory before and when you studied
determinant, one of the first properties that you will see is that det
(A B) is equal to determinant of A times determinant of B. This is
a property of determinant so this is a well-known property of
determinants. So, because this is a well-known property of
determinants, φ is a group homomorphism. So, this is an important
example of a group homomorphism. So, let us look at one more
example. So 5 in my counting.

(Refer Slide Time: 20:49)

So, let’s say G is an arbitrary group. So, I am going to take any

group and fix an element A in G, fix an element A in G. Consider
the following function from the group of integers to G, which
sends φ of A to sorry φ of now because I have used A for the group
element, let me use n for the integer φ of n to be, I define it to be,
A power n. Remember again A power n stands for a*a*a…n times.

So, I am applying the binary operation of our group G to A, n

times so, this is φ of n equals A power n. Now is this a group
homomorphism? So, again remember the definition is φ of m+n
because the group Z is under addition we have to check φ of m+n
should be equal to φ of m times φ of n because that is the group
operation on G. My notation is just φ m times φ n. So, I do not
write * unless I want emphasize the point here. So, is this true?
But what is φ of m+n? This is A power m+n by definition of phi,
m+n goes to A power m+n. This is true and what is φ of m times φ
of n?

72
This is A power m times A power n. Okay, so now the question is
this true? Is A power m+n equal to A power n dot A power m and
the answer is yes, this is true and this came up in some of the work
that we have done in the past when we worked out some
examples or looked at some properties of multiplication. So, I am
not going to spend a lot of time doing this but quickly I will do
this. What is this?

This is a*a*a…. m+n times and this is first I will do a*a …m

times and then I do * note that I am using dot and star
interchangeably here. Second we have n times. Okay, so we have
a*a*a, m times and then *n times but this is, if you use
associatively of group operation and expand this out you have to
do first A here and then you take this and do this A power n so this
is remember A power n (An ).

So, this is A power n + or rather I should use I am using the

distributive property the associative property. So, then you have
another a * power n like that and this is n times. So, this is a
n+1*a n+1*n+1. So if you keep doing this one by one you get a
power m+n. Okay, so this is an exercise for you to check this. So,
check this as an exercise. Okay, so I remarked this on this earlier
when I use the exponential notation I can use the usual properties
of exponents that you have studied about real numbers and
integers and rational numbers. So, this will go through. Okay, so
this is a group homomorphism. So, let us do one more example
and then we will study some properties.

(Refer Slide Time: 25:19)

Now fix again a group but assume now that it is abelian. So, what
is an abelian group? Recall G is abelian if ab equals ba for all a, b
in G. Okay, so any two elements in G commute with each other.

73
So, this is an abelian group. Okay, so now I consider the map φ
from G to G which sends A to A squared. Okay, so what I am
doing? I am taking an element A and I am simply squaring it. Is
this is a group homomorphism?

So, we have to check remember check φ of ab should equal φ of a

times φ of b. Let’s just write both sides of this equation. What is φ
of ab? This is ab squared right? That is the definition send any
element to its square and what is this? This is a squared times b
squared, φ of ab is ab whole squared. Φ of a times phi of b is a
squared times b squared. Now are these two equal? And here is
where we have to use the abelian hypothesis.

So, let us just write it down so ab whole squared is ab ab, right,

this is the definition of ab whole squared ab then ab because the
group is associative so I am going to just remove the brackets and
write this as ab ab. What is a squared b squared? This is a, a,
because that is what a squared is times bb. So, again I will remove
brackets so this is a a b b. See there is no reason in general that
these two are equal. In general these are not equal.

But note that we are in an abelian group. I am starting with an

abelian group. So, aabb I can interchange in any way I want. So, I
can interchange the middle two elements that is abab. So, if I
interchange this because ab equals ba I have noted that ab equals
ba for all the ab, ab equals ba so this is ab times ab. So, this is true.
So, the map which stands a to a squared is a group homomorphism
if the group is abelian. So, now the question naturally arises:

(Refer Slide Time: 28:33)

Is φ a group homomorphism when G is not abelian? Okay is this is

a group homomorphism when G is not abelian. Note that our

74
proof here suggests that it need not be because in general we may
not be able to interchange A & B here. But unless we get an
example we can’t be sure. Just because the proof that we are
trying to give here does not work in the non abelian case. The
statement itself is wrong for non abelian case, you cannot say that
maybe there is some other proof. Okay, so unless you give an
example where this is not a group homomorphism when G is not
abelian you cannot conclusively settle this question.

So, in order to understand this think of the groups that you know
that are not abelian. The simplest such group is S3, so consider
remember that in an exercise in an earlier video we have shown
that any group of order up to 5 is abelian. So, 1, 2, 3, 4, 5 is
abelian. And hence G3 S3 which has 6 elements is the smallest
non abelian group and in the notation that I introduced way back
in one of the first videos, I used these letters to denote the
elements here. These are all bijections of the set {1, 2, 3} to itself.

S3 is always the set of bijection from {1, 2, 3} to {1, 2, 3} so, now

consider the map from S3 to S3, I map FI φ sends FI to FI
squared, for every I from 1 to 6. Right, so 1 goes, F1 goes to F1
squared, F2 goes to F2 squared and so on. So, now I want you to
check as an exercise here and this is a straight forward exercise it
is a good way to again remind yourselves about the group
operation of S3, φ is not a group homomorphism.

Okay, so you can check that φ is not a group homomorphism. For

example maybe I will just quickly tell you how to check this. If
you recall F2, so I will recall for you from the first video. F2 sends
1 to 2, 2 to 1, 3 to 3 and F3 sends 1 to 3, 2 to 2, and 3 to 1. So,
where does F2, F3 go? So let us do this. If you multiply F2 and F3
1 goes to first you send 1 to 3 then 3 to 3, 1 goes to 3, 2 goes to 2
and then 2 goes to 1, 3 goes to 1 and 1 goes to 1. So, is this clear?
So, F2 was this, F3 was this and F2 F3 is this and this if you recall
again the notation that I developed in the video where I discussed

75
this in detail this is same as 1 goes to 3, 2 goes 1, 3 goes to 2. So,
in this notation is F6 because F2, F3 is F6. So, now if this is a
group homomorphism we want

(Refer Slide Time: 32:27)

Where does F2 go under φ now this is F2 squared. We checked

that F2 is actually in order two element in, when we discussed
orders of elements of a group. So, F2 squared is F1 similarly F3 is
also in order two element. So, φ of F3 is F3 squared and that is 1
namely F1, the identity element. Okay, in other words φ of F2
times F3 is φ of F2 F3 is F6 φ of F6 is F6 squared and F6 is an
order 3 element of S3. So, this is not F1. On the other hand, φ of
F2 times φ of F3 is F1 times F1 which is F1. Okay, so φ of F2 F3
is not equal to φ of F2 φ of F3.

Okay, so I have actually checked all the details here. You have a
nonabelian group where the multiplication and squaring is not a
group homomorphism. So, this exercise is completely checked.
However if the group is abelian you have that is a group
homomorphism. Okay, so I will stop the video now, this video. In
the next video we are going to study some basic properties of
group homomorphisms and learn more about subgroups attached
to group homomorphisms. Thank you.

76
 NPTEL

NPTEL ONLINE COURSE

Introductions to abstract

Group Theory

Module 02

Lecture 11-“Properties of homomorphism”

PROF-KRISHNA HANUMANTHU

CHENNAI MATHEMATICAL INSTITUTE

So let’s continue our study of group homomorphisms. Last video

we defined group homomorphisms and looked at various
examples of group homomorphism. I am going to study further
now. I am going to start with some basic properties of group
homomorphisms.

(Refer Slide Time: 00:34)

So properties of group homomorphisms, okay, so some of the

most important properties are the following. So maybe I will write
this as a definition, proposition rather. Proposition: Let φ from G
to G prime be a group homomorphism, it sends, in other words, φ
of AB is equal to φ of A times φ of B for all AB in G. Then we
have two statements φ of EG is EG prime, so just to clarify here E
always stands remember for our identity element in a group.

But here I have two groups G and G prime, so I am going be

denoting which identity I am talking about by looking at the

77
subscript, E sub G is the identity element of G, E sub G prime is
the identity element of G prime, so what I am saying is that in a
group homomorphism, the identity element of the first group goes
to the identity elements of the second group, further we have if A
belongs to G, if A is an arbitrary element of G then φ of A inverse,
that means I first take the inverse of A and apply φ to it, I get the
same answer as first taking image of A and taking the inverse
image, okay, so just ready this carefully, I first take inverse image,
sorry I first take the inverse then take φ or I first take the image
and then take the inverse image.

(Refer Slide Time: 02:48)

I get the same answer so inverse and then φ is same as first φ and
then the inverse, so these are the properties of group
homomorphisms. So let us prove this, this is very easy, the
condition that φ of AB equals φ A times φ of B guarantees this,
that is a powerful condition that guarantees this. Note that EG
times EG is EG, right, this is in G, this operation is taking place,
this is an equation in G, that is because anything commutes with
EG, so in particular anything when you multiply by EG you get it
back so in particular EG times EG is EG, so apply φ to both sides,
okay, now because φ is a group homomorphism, φ of EG times φ
of EG is φ of EG times φ of EG, φ of EG times EG is equal to φ of
EG, I am not changing the right hand side here, φ of EG is same
as, I am keeping it as φ of EG.

But the left hand side becomes this, because of the group
homomorphism property, okay, now let’s look at this, this is in G
prime this is an equation in G prime, we started with an equation
in G applied φ to it and translated completely to G prime so now
what does this mean? You have two elements, so you have an

78
element actually in G prime that you when multiply you get it
back, so we trying to show that φ of EG is EG prime.

But whatever it is, I can multiply by, both sides by, both sides of
this by, the inverse of φ of EG. What do I get? This is an element
of the group G prime, so we have two elements multiplying to this
element, they all happen to be same element, and you can
multiply by the inverse of this, so φ of EG inverse times φ of EG
times φ of EG is φ of EG inverse times φ of EG, correct? So that
is what we have when you multiply out you get this.

But then what is, what is this, because again the group is
associative we can combine these two and that will cancel because
that is inverse of this, so we get, φ of EG on the left hand side and
what is φ of EG inverses times φ of EG this is an element inverse
times itself. So this is nothing but EG prime, this is a property of
inverse in a group, φ of EG is some element, I am multiplying by
its inverse this is some element in G prime, I am multiplying by
it’s inverse so I get EG prime so this proves the first property that I
said, so the identity element of group G maps to the identity
element of the group G prime, right. Second property is similar
and easy, so let us say that we have we have already shown that so
we know that A time so let A, B in G then A times A inverse is EG
right, this is the definition of inverse so apply φ to both sides so φ
of A time A inverse is φ of EG by part one it is already shown to
be EG prime φ of EG is EG prime, now the group homomorphism
property says that φ of A times φ of A inverse is EG prime.

(Refer Slide Time: 06:50)

So φ of A times φ of A inverse is EG prime, but remember inverse

is the unique element which has the property that φ of A times it is
EG prime, so by definition of inverse, we must have φ of A

79
inverse is φ of A inverse the whole inverse because φ of A when
you multiply with this element here you get the identity element
so this element must be the inverse of φ of A which is denoted by
φ A whole inverse.

So this is, remember, the second property, φ of A inverse is φ of A

whole inverse so this proves two, so the proof is complete, in
other words, identity must go to identity under a group
homomorphism and inverses must go to inverses, this is useful to
keep in mind because it tells you quickly if a map is a group
homomorphism or not, if it doesn’t send identity element to
identity element, it cannot be a group homomorphism.

So, if you recall, one of the examples I did in the previous video
was sending a function from Z to {1, -1}, φ of A is, let us say 1, if
A is odd, -1 if A is even, and I asked you to check that this is not a
group homomorphism. That was left as an exercise for you. Now
let’s do this exercise using the proposition that we proved today.
Remember the other way if you send even numbers to 1, odd
numbers to -1, that was a group homomorphism.

But if I interchange this it is not a group homomorphism. In fact,

what is the identity element of Z? This is my notation right, E is
always the identity element and I am denoting the group by the
subscript. Remember Z is a group under addition so the identity
element is 0. Right, and zero is even, so under this map, φ of 0 is
because, even numbers goes to -1, φ of 0 is -1, but what is, you
call this group G prime, what is EG prime?

Certainly 1, because this is a group under multiplication with 1 as

identity, so the zero element does not map to the identity element.
So, φ of EG is Z rather is not equal to EG prime. So, φ is not a
group homomorphism. We do not need to check anything more,
you simply send, you simply check that the identity element does

80
not go to the identity element it cannot be a group
homomorphism. Okay, so similarly inverses go to inverses.

So however I want to make a point here, if the identity element

goes to the identity element does not mean that it is a group
homomorphism. The proposition does not say that if this condition
is satisfied if EG goes to EG prime, it is a group homomorphism.
It only says that if you have a group homomorphism,

(Refer Slide Time: 10:55)

identity element goes to identity element. So, neither of these are

sufficient conditions for homomorphism. There only necessary
conditions. They must be true, but if they are true it does not mean
that it’s a group homomorphism.

Okay, so now that we have this, let me define some important

subgroups associated to, important subgroups associated to a
group homomorphism. Okay, so there are two.

So let me work now with an arbitrary group homomorphism. Let

us say φ from G to G prime is a group homomorphism. So we
define two subsets first and we will check that they are group
homomorphisms, sorry we define two subsets first and check that
they are subgroups. So “kernel of φ”, okay this is the word, kernel
of φ and it is denoted simply by Ker (φ), is the following. So, you
define Ker (φ) to be all elements of the group such that φ of A is
the identity element of G prime. So this is all elements of the
group such that φ of A is EG prime, exactly the subset of elements
which map to EG prime, so and similarly let me define this first.
Image of phi, so this is the subset of G remember. Kernel is a
subset of G. Image φ

(Refer Slide Time: 13:14)

81
which I will denote by image phi is the following. Image phi is
equal to, simply the image of the function φ, so this is just a set-
theoretic notion, I am taking all elements of G prime, this is inside
G prime clearly. All elements of G prime which appear as the
images of the elements of G under φ. So, the proposition now is
kernel φ is a subgroup of G, it is a subset of G by definition but in
fact it is a subgroup of G and image φ is a subgroup of G prime.

This is true for any group homomorphism. So if you start with any
group homomorphism φ, kernel is a subgroup of G, image is a
subgroup of G prime. What is a proof? It is again fairly straight
forward, let me check 1. What is a subgroup? If you recall from a
previous video, subgroup is a subset of the group which is closed
under multiplication, which has inverses and which has the
identity element, Okay.

So first of all is it closed under, kernel φ is closed under the binary

operation of G. So, let’s check this. Kernel φ is closed under the
binary operation of sorry G not G prime of G because if, A and B
are in kernel phi, this means φ of A and φ of B are both EG prime,
by definition. Kernel φ consists of elements which map to the
identity of G prime, but then what is φ AB? This is by definition
sorry not by definition by the property of a group homomorphism,
is same as φ of A times φ of B which is EG prime times EG prime
which is EG prime. So that means AB is in, right, because AB
maps to EG prime so AB is in the kernel phi. We started with two
elements in the kernel and concluded that their product is in the
kernel. Kernel φ contains EG. This is very easy, right, because
what is φ of EG? This is by the previous proposition this is EG
prime. So that means EG is in the kernel φ.

Finally kernel φ is closed under inverses, the third property, that

we require for a group, subgroup, is it closed under inverses? Yes,
because if a belongs to the kernel φ that means φ of a is EG prime,
by definition it is EG prime. By the previous proposition, φ of a

82
inverse is φ of a inverse by the previous proposition, right and
this, sorry, so let me write it like this,

(Refer Slide Time: 16:59)

Φ of a inverse which, by the previous proposition, is φ of a inverse

which is EG prime inverse which is EG prime, right. φ of a
inverse is by the previous proposition inverse of φ a, φ of a is EG
prime, it is EG prime inverse, inverse of identity itself okay, so
that means a inverse is in kernel phi. So we have checked three
properties: kernel φ is closed under binary operation of G, it
contains the identity element and it contains inverses, so kernel φ
is a subgroup of G.

Now let us check that image is the subgroup of G. I am going to

show that image is of subgroup of G. Similarly, exactly as before
image φ is closed under binary operation of G prime, right, now
we are claiming that image φ is a subgroup of G prime, so it must
be closed under the binary operation of G prime. What are two
elements of image φ? If you recall, image φ is φ of a, a in g, for all
elements a of G, we take φ of a, so two elements of image φ
would be φ of a times φ of b. Let’s say they are both in image φ.
But that means, φ of a times φ of b, by the property of group
homomorphism, is same as φ of ab, but this is by definition in
image φ, because this is the image of ab, product of φ of a and φ
of b is the image of ab.

So if you start with two things in image φ, their product is in

image φ. Image φ contains EG prime, it must contain EG prime in
order to be a subgroup but it does, because φ of EG by the
previous proposition is EG prime. So EG prime is in the image,
recall again that image is the image of the function, EG prime is
an image of an element. So it is contained in the image, it is the set
theoretic image. Image φ is closed under inverses, this is also easy,

83
let’s say φ of a belongs to image φ. Then what is φ of a, whole
inverse? By the previous proportion, this is same as φ of a inverse.

(Refer Slide Time: 20:01)

Right, φ of a, whole inverse is φ of a inverse, but φ of a inverse

again is by definition an element of the image. Because it is image
of a inverse, so image φ is a subgroup and this proves the
proposition. So we have shown that kernel and image of
subgroups G and G prime, so this is an important, these are two
important subgroups attached to any group homomorphism. So it’s
a good exercise now for you, to work out the kernel and images in
all the previous examples that we discussed of group
homomorphisms. I won’t discuss all the examples, but if you
consider the map from Z to Z, φ of a being na, so we have fixed n,
this is the first example of a group homomorphism that we
studied, you fix an integer n and send an integer a to n times a, the
kernel of φ is all integers such that na is zero, because zero is the
identity element of Z.

But this is precisely zero, so the kernel is just zero. Remember

kernel, being the subgroup of the group, contains the zero
elements always, so in this case it is exactly zero element. So this
is a sub group of the first Z, what is the image of Z, phi? This is all
na, where a is in z, and this is precisely the group nZ that we
discussed, subgroup of Z obtained by multiples of a, so this is a
subgroup.

And if you take the determinant homomorphism from GLNR to R

star, what is the kernel of this? Remember φ of A is determinate of
A, kernel of a is the set of matrices in GLNR, such that φ of A
which is the determinate of A, is 1. So this is a subgroup by the
previous proposition and this is called the “Special Linear Group”

84
denoted SLNR, okay, Glnr is called the “General Linear Group”.
Special linear group is a group of invertible elements with
determinant 1. That’s the kernel. What is the image of φ?

(Refer Slide Time: 23:31)

Image of φ is the set of real numbers which are obtained as

determinant of A, for an invertible matrix. So image consists of all
determinants as you vary the invertible matrix in Glnr, what is
this? It’s certainly a subgroup of, by the previous proposition, it is
a subgroup of the multiplicative group of nonzero real numbers.
But I claim in fact, it is all of nonzero real numbers, why?

I claim that image of the determinant map is all of nonzero real

numbers. In other words, given a nonzero real number, r let’s say,
there exists an invertible real n by n matrix A such that
determinant of A is r. Note that if I show this, then I have justified
saying that image of φ is R star, because if there is such a matrix,
it is in Glnr, so A is in Glnr, that is to say it is an invertible real n
by n matrix and its determinant is r. So r would be in the image, so
if I want to show image of φ is R star I need to verify this, but
what is such a matrix? I simply can take, I will take a diagonal
matrix, I will put r in the first left hand side position and I will
take ones everywhere, I will take one everywhere else on the
diagonal, zeros off the diagonal. This is certainly an invertible
matrix because its determinant is r and r is non zero. It is an
invertible matrix, it is size N by N, it has real entries. So this
matrix will have determinant because determinant of a diagonal
matrix is the product of diagonal entries, determinant of A is r. So
the image of the determinant homomorphism is all real numbers,
non-zero real numbers of course, we want non-zero real numbers.

(Refer Slide Time: 26:03)

85
Just one more example I want to do, this is also something we
discussed in the previous video. If you take, G is an arbitrary
group, fix an element of G and consider group homomorphism
from the integers to the group by sending n to An this is a group
homomorphism was something we checked earlier, what is the
kernel of φ? This is all integers such that φ of A, in this case An is
identity of G. This is a subgroup of Z. This is a subgroup of Z, so
this is all integers such that A power n identity is a subgroup of Z.
What is the image of φ? This is An so remember this is φ power n
as n varies over integers, this is by definition image of φ, but what
is φ power n? This is just A power n has n varies over integers. In
other words, this is just A-3, A-2, A-1, A0 which is EG, A, A2 if
you recall and you remember now from a previous video we gave
a name for this group.

This is simply the subgroup generated by A, this is the subgroup

generated by image of φ is the subgroup generated by A. Let’s
look at kernel slightly more carefully, so now what is kernel of the
same example, in this same example what is the kernel of φ? So
kernel of φ, as I wrote earlier, is all integers such that A power n
equals EG and if you recall again from a previous video we have
already classified all the subgroups of the integers.

(Refer Slide Time: 28:38)

What are they? Recall we earlier classified, so remember this is a

subgroup of Z, we already classified all subgroups of, what are
they? They are of the form, they are multiples of a fixed integer,
so we have already used n and a here, so I am going to use BZ,
right, they are of the form BZ. BZ, being all multiples of a fixed
B, so BZ is BN, where N varies over, so they are of the form BZ,
where BZ is this, so every subgroup of integers is like this.

In particular, kernel is also like this, so in our example kernel φ

being a subgroup of Z is of the form kernel φ is BZ for some

86
integer, in fact, B is a non-negative integer. See if you go back and
see the video where classified subgroups of Z they are of the form
BZ, for a non-negative B. We can always if, ok, I don’t want to
repeat the proof, but we first rule out the case that it is a zero
subgroup, in which case it is of the form zero times Z.

Otherwise there are some numbers in it, nonzero numbers, hence

there must be some positive number in it and we take the smallest
positive number and show that all elements of this subgroup are
multiples of it, so B is either 0 or positive. So now this B is
actually something that you are familiar with, B as an exercise
note that B is nothing but the order of, recall that, from before
what is order of A? So we have two possibilities.

So what is order of A? We first look at multiple of A, so we look at

E, A, A squared, A cubed, we keep looking at this, either you will
never hit E again the identity element or you will hit E
somewhere. It is the first place you hit E if you hit E. So order of
A has two possibilities, it is either, okay, maybe I should write like
this, if A power n is e for some positive integer,

(Refer Slide Time: 32:38)

n, then order of A is the smallest positive integer m such that A

power m is E, this is such that. Okay, so S.T. always stands for
“such that”. So if A power n is identity for some positive integer n,
I am going to simply look at the least positive integer which has
that property. If A power n is not equal to E for any positive
integer n then I will simply define the order of A to be infinity, and
now I claim that looking at the map, the group homomorphism
which sends integer n, A power n in a group G, we noted that
kernel is of the form BZ, it is all multiples of B, so kernel of
remember φ, is all integers n such that A power n is identity. And
we saw that it is BZ, for some integer which is non-negative. So
it’s either 0, in which case kernel φ is zero, kernel φ is the zero set

87
and in this case A power n cannot be equal to A for any positive
integer, so order of A is infinity.

So actually I should not say this, so this is not right, B is nothing

but the order of A is not quite true. B is related , I should write B is
related to order of A, if B is 0 then order of A is infinity. If B is not
0, so B is positive, then kernel of φ is BZ and now the order of A
is actually B. So in this case order of A is B because remember
kernel is BZ and B is positive, so the smallest positive integer
such that An is B, so order of A is B.

So if the kernel of this map is kernel of this map is zero if kernel

of this map is zero then order of A is infinity and kernel of this
map is nonzero then order is equal to the generator of the group,
so B is related to order of A. So these are some of the properties of
group homomorphisms and I am going to stop here in this video
and in the next video I am going to talk about, so in this video we
have looked at basic properties of group homomorphisms, namely
that they send identity element to the identity element, inverses to
inverses, we looked at two very important groups associated to a
group homomorphisms, namely the kernel and the image and we
worked out what these are in some of the examples of group
homomorphisms that we looked at. In the next video I will
continue my study of group homomorphisms and define
isomorphism of groups. Thank you.

88
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 03
Lecture 12 – “Group isomorphisms”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHAMETICAL INSTITUTE

Let us continue our discussion about group homomorphisms. We defined kernels

and images of group homomorphism in the previous video, so let me do one
immediate proposition following those definitions.
(Refer Slide Time: 03:06)

So, before that let me recall define some well known terms. So let us say φ is a
group homomorphism from G to G prime, so let’s say φ is a group homomorphism
from G to G prime. I define φ is 1-1 or injective is also the word, a word that we
use, injective if φ of a, b is equal to phi of b implies a equals b, okay this is one
way of writing this equivalently, if A is not equal to B then phi of A is not equal to
φ of B, okay so this is actually nothing to do with a group homomorphism, this is a
purely set-theoretic notion.
So it’s injective means two distinct elements map to two distinct elements.
Similarly φ is onto or another word for this is surjective if image phi, my notation
was small im, okay these is also purely set-theoretic notion, image of the map is all
of G prime. So this is very simple, but the proportion I want to do before
continuing today is that φ is 1-1 if and only if kernel of φ is just the identity
element of the group.
So remember kernel is the set of elements of the group G which map to the
identity element of group G prime so 1-1-ness can be captured purely by looking at
kernel of this, so what is the proof of this? So suppose we have to prove two
directions right, φ is 1-1 if and only if kernel φ is eg, so let us suppose φ 1-1 let us
prove that kernel φ is eg and on the other hand let suppose kernel φ is eg and to
that the φ is 1-1.
So suppose first that φ is 1-1 and let A be in kernel of φ, then by definition φ of A
is eg right, kernel of phi by definition elements which map to eg, but we have

89
already seen that φ of eg is same as, sorry φ of eg prime right, φ of it is the
identical element of G prime A being in the kernel means, A maps to eg prime.
But eg also a maps eg prime that’s a property of the group homomorphism, but
now φ is 1-1 and φ of A is equal to φ of eg so A equals eg. Right, if A is in the
kernel A must be eg, certainly eg is in the kernel that’s always true. Because kernel
φ is a subgroup it contains the identity element or more directly identity element of
the group maps to the identity element of the group G prime so kernel contains eg
and if anything else is contained in eg sorry if anything else is contained kernel φ it
is equal to eg so kernel φ is simply eg.
So that we have shown, so if kernel φ is 1-1 then the kernel of φ is eg. So the next
direction,
(Refer Slide Time: 04:42)

this is the easy direction, so suppose that kernel of φ is eg, so we have now
assumed that kernel of phi is eg, kernel of φ eg. So we want to show that φ is 1-1,
so suppose what is 1-1? it means that if two things map two, if two element map to
the same element they are equal. So suppose φ a is equal to φ of b, okay, if φ of a
is equal to φ of b then φ of a times φ of b inverse equals identity of G prime,
because φ of a, I am simply multiplying by φ b whole inverse.
So this is happening in G prime, φ of a times equal to φ of b, φ a φ b inverse is eg
prime, but φ of b inverse remember by the properties of group homomorphism that
we studied earlier, φ of b inverse is nothing but φ of b inverse φ of b whole
inverse, when I write it like this I mean φ of b inverse, that is same of φ of b
inverse. But again using the property of a group homomorphism this is just φ of ab
inverse, φ of a times φ of b inverse is φ ab inverse and this is φ of, okay now φ of
ab inverse is in eg that means, ab inverse is in the kernel φ the definition in kernel
φ is that it consists of those elements which map to the identity element of G
prime.
So ab inverse is in the kernel phi. But note that kernel φ is precisely eg, we have
only one element in kernel phi, so that means ab inverse is eg, because eg is the
only element in kernel φ, so ab inverse must be in the eg, but that means a is equal
to b, because we can multiply by b. So ab inverse is eg means a equals b, so we
have stated with φ of a is equal to the φ of b and concluded that a equal to b. That
means, so φ is 1-1, okay so checking if a group homomorphism is injective or not,
in general you have to check that no two distinct element go to distinct element,

90
but the advantage with a group homomorphism is that all you need to check is the
kernel, if kernel consist only of the identity element, it is automatically a 1-1 group
homomorphism.
Okay, so a group homomorphism is 1-1 precisely if kernel is the identity, or kernel
is just the identity element. Now a very important definition.
(Refer Slide Time: 07:50)

An “isomorphism of groups” is a group homomorphism, I am introducing another

word for you, an isomorphism of groups is a group homomorphism φ from G to G
prime such that φ is 1-1 and onto, so a group homomorphism which is
simultaneously 1-1 and onto is a group isomorphism. Okay, so an immediate, so
isomorphism of groups is a very very important concept of groups, it says that two
groups are isomorphic when they are essentially same, if there two groups are
isomorphic we will consider them as same groups.
As far as the group theoretic properties are concerned there is no difference
between them, because there is a bijection between them so this is simply saying
that, remember I used the word bijective earlier, bijective simply means its 1-1 and
onto, so because it is bijective the sets are same and because it is homomorphism
multiply two things here in one group is same as multiplying two things in the
other group, so as far as the group theoretic properties are concerned they are
same.
So an exercise for you which I will not discuss in great detail, if φ from G to G
prime is a group isomorphism, then the map φ inverse, remember that if it is a
isomorphism is a bijective map, φ is a bijective map so there is a map which sends
G prime to G namely φ inverse just the set theoretic inverse function, so where
does an element under G prime go under φ inverse , you simply look at a pre-
image because there is exactly one pre-image it is well-defined.

The map φ inverse is also a group isomorphism, if φ is a bijective map since φ is a

bijective map φ inverse can be defined. Right, that is because if b is in G prime
define φ inverse b equals a if φ a is b. So you have G here G prime here, a goes to
b under φ right, because it is a bijective map b has a pre image and it is exactly
one, there is exactly one pre image, so I simply send b to a under φ inverse. Okay,
so if a goes to b under φ, b goes to a under φ inverse.
So you can define it, because it is a bijective map, otherwise you can’t define the

91
inverse. And it is obviously a bijective map also, because if φ is bijective its
inverse image is not only defined but it is bijective. So its bijective map, to check
that it is a group of isomorphism we have to check that, φ inverse is actually a
group homomorphism and this I will leave for you to check, so you have again in
the picture I drew earlier, you have two elements b1 b2 lets say, which map to a1
a2 and you also have I want to draw a bigger diagram.
So you have b1 b2 and b1 b2 okay that’s in the G prime. You have φ of this is a1,
this is a2 and this is a1 a2, a1 maps to b1 under φ inverse it maps back to a1.
Similarly a1 a2 map to each other under φ and φ inverse, but because φ is a group
homomorphism a1 a2 maps to b1 b2 that is the property of group homomorphisms,
but again by inverse definition of the inverse function this must map to b1 b2 must
map to a1 a2.
So this gives you an idea of how to check that φ inverse is a group is a group
homomorphism, so if you have an isomorphism of a groups which is by definition
a group homomorphism which is bijective, you have an isomorphism in the
opposite direction.
So group isomorphism is an important property because it allows us to consider
two groups as same if their group theoretic properties are same, we do not want to
think them as a different just because the names of elements are different or the
way which it is represented is different. Okay, so we will as the course progresses
we will look at various examples of this, but just to give you very basic idea, I
want to consider one very simple example of group isomorphism,
(Refer Slide Time: 14:33)

or few of them , so let’s take G1 to be 1, I, minus 1, minus I, so where I was a

square root of minus 1, so remember this we discussed this came up earlier in our
videos, this is the group consisting of fourth roots of unit. So this is a subgroup of
C star. On the other hand let us consider G2 to be 1, a, a squared a cubed, where a
is just a symbol okay, meaning it has no further interpretation, it is just a symbol, it
is not a number or a matrix or an operation or a function, or permutation and so on,
it’s just a symbol.
And the property is, the property that it satisfies is, okay let me use e for the
identity element, a to the fourth is e, so now G2 is a group, it is a group because it
has identity, it is closed under the multiplication because a, a squared is a cubed is
a square a cubed it is, what is a square a cubed? Just the exponential rules tell you

92
it is a5, but a5 is a4 times a, right, and a4 is e so this is just e. So essentially what
we are doing is exponents so similarly a cube power four, what is this?
Okay, this is a to the 12, and that it is a to the four cube, that’s also e, a cubed to
the third that’s a to the ninth, so that is a to the 8 times a so that is a. Okay, so using
this very important property this symbol satisfies, we can check that, this is closed,
these are just examples but in general any a power i is inside {e, a, a 2, a3} and e is a
power zero. Okay, because you can basically divide i by four and the reminder is
what you get here, because a power four is e, so G2 is a group, it is an abstract
group in the sense that a is just a symbol with no meaning attached to it, G1 is also
a group but here there is a concrete meaning I is a square root of minus one I is a
complex number. I claim that G1 and G2 are isomorphic, why is this? Let’s define
a function so remember an isomorphism is, when I say isomorphic so the word
isomorphic,
(Refer Slide Time: 17:57)

We say G1 and G2 are isomorphic,

(Refer Slide Time: 18:13)

if there is a group isomorphism okay this is just a notation. We say that two
groups are isomorphic if there is a group isomorphism from one to the other. Okay,
in this now I come back to this example I want to say G1, G2 are isomorphic. So
we have to exhibit an isomorphism, so let’s define an isomorphism, so G1 is 1,e, 1,
i, minus 1 , minus I, so send 1 to 1, I want to do from G1 to G2, I send 1 to e, I to a,
minus 1 to a square, minus I to a cube. Okay and one can check easily, this is a
group isomorphism, it is certainly a bijective function right, because four elements
here go to four distinct element here, G2 has four elements and everything is in the
image and no two elements map to same element, so it is a bisjction.
That is easy, it’s just from the definition. It is also more or less easy and it is a
group homomorphism, it is also easy to check this, let me not write all the details
but just say how to check this, for example where does I squared go? What is I
squared? Let us check that, φ of I squared is φ of I times φ of I. That is what a
group homomorphism is supposed to do. But φ of I squared is φ of minus one,
because I squared is minus one, what is φ of minus one, in my notation it is, in my

93
definition it goes to I squared, on the other hand what is φ of I? It is I, this is an φ
of I is a, it is a times a and that is a squared, this is okay.
I just check one more just for your clarity, what is φ of minus one times minus I,
this is φ of, so I am multiplying these two elements minus one and minus I, that is
φ of I. right, φ of I is a, what is φ of minus one times φ of minus I? Φ of minus one
is a squared, φ of minus I is a cubed, a squared times a cube is a, so that I checked
earlier. So this is also okay, this can be checked to be a group isomorphism.
So that is not difficult G1, G2 are isomorphic groups, but I want you to think
about this carefully, G1 is a concrete group it is a subgroup of complex numbers
non zero complex numbers which is fourth roots of unity, so it has a some meaning
complex numbers have some other structure, they have some geometric meaning
and so on.
Whereas second group has no structure, no meaning, other than whatever the group
axioms give it. a is just a symbol and a4 is e, but G1 and G2 are isomorphic
groups, so in abstract group theory, we are only interested in the group theoretic
properties of a set. So in the study of abstract groups G1 and G2 are considered
same, there is no difference between them, all the extra structure the G1 has is
irrelevant as far as the group theory concerned, G1,G2 isomorphic as groups which
is to say they essentially same.
There is no difference between them as groups,
(Refer Slide Time: 22:30)

Also note that, G2 is cyclic, G2 is a cyclic group. It is, in fact G1 is also a cyclic
group, so cyclic group remember means, there is an element which generates the
group in G1 case I generates the group, I squared is minus one, I cubed it is minus
I, so G1 is cyclic and in G2 a generates it. So if G1 and G2 are isomorphic and
both are cyclic. So it is not surprising there is in exercise I will give you and I will
let you do this on your own.
If G1 is, now this is not G1 and G2, let us says G and G prime are isomorphic,
then so there are two exercises G is abelian if and only if G prime is abelian, G is
cyclic if and only if G prime is cyclic. Okay, being cyclic or being abelian is a very
group theoretic property, so if earlier my statement that if isomorphic groups are to
be thought of as same, certainly I would expect that abelianness would be care
preserved under isomorphism.
So if one is abelian other must be abelian, otherwise I cannot think about them as

94
two groups, because abelianness is necessary property of a group if one group is
abelian and another group is not abelian I cannot think of them as same groups.
Similarly if one group is cyclic and another is not cyclic, I cannot think them as
cyclic, I cannot think them as same, so if two groups isomorphic, one is cyclic if in
and only other is cyclic, one is abelian if and only other is abelian.
So again I should remark that this symbol here means, if and only if, I have been
using this perhaps without saying what it is, its if and only if. That means G is
abelian if G prime is abelian and only if G prime is abelian. So in order to prove
this, you want to show that if G is abelian then G prime is abelian, same similarly
you want to show that if G prime is abelian then G is abelian. Okay, so I will not
do this exercise for now, because it’s instructive to do this on your own and it is
not difficult.
Online Editing and Post Production

Online Editing and Post Production

Karthik
Ravichandran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh Kumar
Web Studio Team
Anitha
Bharathi
Clifford
Deepthi
Dhivya
Divya
Gayathri
Gokulsekar
Halid
Heamvathy
Jagadeeshwaran
Jayanthi

95
 Kamala
Lakshmipriya
Libin
Madhu
Maria Neeta
Mohana
Mohana Sundari
Muralikrishnan
Nivetha
Parkavi
Poonkuzhale
Poornika
Premkumar
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirley
Subash
Suriyaprakash
Vinothini
Executive producer
Kannan Krishnamurty
NPTEL Co-ordinators
Prof.Andrew Thangaraj
Prof. Parthap Haridoss
IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resource Development
Government of India
HYPERLINK "http://www.nptel.ac.in" www.nptel.ac.in
Copyright Reserved

96
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 03
Lecture 13-“Normal subgroups”
PROF.KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE
(Refer Slide Time: 00:16)

Okay, so with that let me define an important concept of normal subgroups,

(Refer Slide Time: 00:23)

which is related to both the notion of subgroups and group homomorphisms. So the
definition is, let G be a group, a normal subgroup H of G is a subgroup, so I am putting
an adjective before subgroup, it is normal subgroup, so first it must be a subgroup, and it
must satisfy, more conditions. A normal subgroup H of G is a subgroup satisfying the
following property, so it is a subgroup that satisfies the given g in G, and h in H, ghg-1 is
in G, sorry it is in H, so given an element of the group and an element of the subgroup, I
must have that, g h g-1, so the group element times the subgroup element times the inverse
of the group element, must again be an element of H, okay so this is what we call a
normal subgroup. So an immediate example. If G is abelian, then every subgroup is
normal, okay this is easy to check because, a normal subgroup is a subgroup, which has
an additional property that ,if g is in G, and h is in H, g h g-1 is in H, the second condition
is automatically true in an abelian group, because if g h g-1 in an abelian group, you can
multiply in any order.
This is same as g g-1h,so I can interchange, so now g g-1h is nothing but h. Remember that
I have to start with h in H, and g in G, because h is in H, ghg invers is in H, so there is no
problem in an abelian group, so normal subgroups, really are interesting if you do not
have an abelian group, if you have an abelian group, every subgroup is normal, it is not
an interesting new concept, and on the other hand,
(Refer Slide Time: 03:26)

if you have a non-abelian group, you can have subgroups , which are not normal, okay
for example, take G to be S3,again remember is the set of bijections from a 3 element set
to itself, and in my notation these are {f1,f2,f3,f4,f5,f6}. If you take the subgroup H to be
{f1,f2}, and I will let you check this, H is a subgroup of S3 , because f1 is identity
element, and f22 is f1, so I will remind you what is f2, f2 is the function which sends 1 to

97
2, 2 to 1, and 3 to 3, and so f22 is f1, which is the identity function, so it is a subgroup,
but H is not normal in S3.
So H is a subgroup consisting of f1and f2, I want to show H is not normal in S3,so let us
take f3, f3 if you recall, is the element , which sends 1 to 3, 2 to2 , and 3 to 1, so let us do,
so remember, a normal subgroup is a subgroup satisfying: take something in the group,
something in the subgroup, and do this, g h g-1 it is in H, so I am taking something in the
subgroup namely f2, and something in the group namely f3, and I want to calculate f3 f2f3-
1
, but f3-1 if you think about it, is just f3, because the f3 has order 2, and f32 is also f1, so f3
is f3-1 .
So what is f3 composed with f2 composed with f3, so let us do this systematically. And
under f3, 1 goes to 3, and under f2, 3 goes to 3, and under f3, 1 goes to 1, right and where
does 2 go ,so this is the f3, this is the f1 and then f3. 2 goes to 2 under f 3 1 under f2, 2
goes 1 under f2 and 1 goes to 3 under f3. Similarly 3 goes to 1 under f3, 1 goes to 2 under
f2 and 2 goes to 2 under f3. So f3f2f3 is the function which sends 1 to 2, 2 to 3 and 3 to 2
right, and this is not equal to f2, right it is not equal to 2 and it is certainly not equal to
f1. So f3f2f3 inverse is not in H which recall, so H is not normal. So it is not a normal
subgroup.
Abelian groups all subgroups of abelian groups are normal but non-abelian groups have
subgroups which are not normal. I want to now introduce two important examples of
normal subgroups.
(Refer Slide Time: 07:47)

Okay, so let us take any group homomorphism: let us say this is a group homomorphism.
Then kernel of phi which we already seen is a subgroup of G is actually normal subgroup
of G, so we already saw kernel is a subgroup. To check normality, what do we need to
check, so let g be in G and let h be in kernel phi. We want to check that g h h inverse is in
kernel phi. What is the meaning of being in kernel if you apply phi to it must send it to
the identity, so let us check that. What is phi of ghg inverse.
(Refer Slide Time: 09:27)

It is because phi is a group homomophisim this is phi g phi h phi g inverse, okay. This is
phi of g what is phi of h because h is in the kernel phi of h is eg’ and phi (g) whole
inverse so eg prime is the identity this is just phi of g inverse which is eg’ so ghg inverse
okay. So this implies kernel phi is a normal subgroup, so kernels are always normal
subgroups, so this is an important example of normal subgroup. Whenever we encounter
a group of homomorphism and take its kernel it is automatically a normal subgroup.
(Refer Slide Time: 10:20)

A second important example of a normal subgroup is the “centre of a group”. Centre of a

group is always a normal subgroup, so recall what is the centre of a group g it is denoted

98
by Z (G) this is all elements g in G such that ag =ga for all, again already saw that Z(G) is
a subgroup. That is not surprising, that is not new to us. To check the normality just like
before we take a group element and a subgroup element and let us check if, what do we
want to check, we want to check that ghg inverse is in Z(G), what is meaning of being
Z(G)? It means that it commutes with everything in the group so let a in G be arbitrary
we want to check that ghg inverse commutes with it so let us write ghg inverse times a,
so I want to rewrite this as gh g inverse a. Now let us use property that h is an element of
the centre, if a h is element of a centre, h commutes with everything so since h is in the
centre let me write here.
h is in Z(g), h commutes with everything, so in particular it commutes with g inverse a,
so h times g inverse a is g inverse a times h. I am not disturbing this g so I am writing this
like this. But now actually I should have been, there is no need to do all this sorry, so it is
much easier than this, because you want to check that gh g inverse Z(G) but because h
commutes with everything.
(Refer Slide Time: 13:30)

So this is g inverse time h, right? h is in the centre, so h g inverse is g inverse h, so I

want to use that, this just h, which is that taken to be in centre. So if g is in the group h is
in centre, ghg inverse is in the centre, okay. Centre is a normal subgroup and kernels are
normal subgroups. These are two very important examples of normal subgroups okay, so
I will stop this video here, in this video we have looked at normal subgroups, and we
have looked at group isomorhipisms okay, and this is, these are both very important
properties and especially normal subgroups will keep coming up a lot in our course and
next time I will start talking about cosets of a group and talk about quotients of groups,
quotient groups, thank you.
Online Editing and Post Production
Karthik
Ravichandran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh Kumar
Web Studio Team
Anitha
Bharathi
Clifford
Deepthi
Dhivya
Divya
Gayathri

99
 Gokulsekar
Halid
Heamvathy
Jagadeeshwaran
Jayanthi
Kamala
Lakshmipriya
Libin
Madhu
Maria Neeta
Mohana
Mohana Sundari
Muralikrishnan
Nivetha
Parkavi
Poonkuzhale
Poornika
Premkumar
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirley
Subash
Suriyaprakash
Vinothini
Executive producer
Kannan Krishnamurty
NPTEL Co-ordinators
Prof.Andrew Thangaraj
Prof. Parthap Haridoss
IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resource Development
Government of India
HYPERLINK "http://www.nptel.ac.in" www.nptel.ac.in
Copyright Reserved

100
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 03
Lecture 14 – “Equivalence relations”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

Okay so far we have studied various properties of groups, subgroups, normal

subgroups, group homomorphisms, and isomorphisms of groups and so on. And
we next want to study an important operation called quotients of groups and in
order to go that I want to talk about cosets of a subgroup in a group. Before that let
me recall for you quickly the motion of equiavalence relation on a set which many
of you may have seen before.
But it is very important for the course, so let me quickly define it and check its
important properties. Then we will apply to the case of a subgroup of a group.
(Refer Slide Time: 01:06)

So review of equivalence relations on a set. So let S be a set okay, for this video
we are only going to look at some sets, there is no group here. So let S be a set and
equivalence relation on S is a relation, okay, it is relation denoted by the symbol
here, it is a relation of among elements of S. So we say that two things are related
and we write this symbol A ~ B. This means that A is related to B, so A is related
to B is denoted by this symbol A ~ B.
So an equivalence relation is a relation on a set, which is to say a relation among
elements of that set, so it is very precisely speaking, it is a subset of a S cross S.
But I do not want get into this formal definition because it is confusing, if you have
not see it before. So it is simply a relation and when you see this you think of a
way of saying that two things are related or it tells you given two elements, if they
are related or they are not related. But the relation must satisfy the following
conditions.
A must be related to A for every element of S, okay an element must be related to
itself. If A is related to B then B is related to A, okay. So this is true for every a,b
in S. The third property is called transitive property, if A is related to B, B is

101
related to C, then A is related to C, so this is true for all a, b, c in S, okay. An
element must be related to itself, if an element is related to another that element
must be related to A. If an the element related to B, B is related to C, okay. The
most important example, a trivial example really, is equality: take for example on
Z okay, you say that a is related to a, if a = a. So equality is a basic example of an
equivalence relation, every element it is equal to itself, if a=b, then b=a, if a=b,
b=c, then a=c. So there is no surprise that equality is an equivalence relation. But
there are more interesting equivalence relations and hence you actually do talk
about equivalence relations because there are equivalence relations that are not
equalities. So another example.
(Refer Slide time: 04:46)

Take again the integers, but now you say that and you say that a is related to b if a-
b is divisible by 4. So this is my definition of a new equivalence relation so this is
the definition. For example 5 is related to 5, of course. Because 5-5 is 0. So earlier
equality was the only equivalence relation but now we have more, because 5 is also
related to 1. And so 5 related to 5 because 4 divides 5-5 which is 0, 5 is related to 1
because 4 devised 5-1 which is 4.
Similarly 5 is related to 21, right, rather, yeah, 21 because 4 divides what is 5-21?
That is -16 right and 4 divides -16, so 5 is related to 25, 5 is related to 29, 5 is
related to 34, or 33 and so on. So 5 is related to 5, but it is also related to a lot of
other integers. But 5 is not related to let me use this symbol to 26, so 5 is not
related to because, what is 5-26? 4 does not divide 5-26 which is -21 so 5 does not
divide, 4 does not divide it. So 5 is not related to 26 so this is a more interesting
example of equivalence relation than equality. So because we are doing group
theory, let me define another equivalence relation,
(Refer Slide Time; 07:18)

which is more appropriate for us and more important for us. So let G be a group
and let H be a subgroup of G. So we have a group and subgroup of that group, so I
define A is related to B, so let me start like this let a, b be two elements of G. We
define A is related to B if let’s say, A inverse B is in H. So actually I forgot
something here, before I continue with the third example, I just wrote something I

102
never said that it is an equivalence relation.
I should prove it right, so that I will not check, I will quickly say why it should be
true. The equivalence relation in the second example is that A is related to B if 4
divides a-b. So certainly, so why is this an equivalence relation. So certainly A is
related to A that is okay because A-A is 0 and 4 divides it. If a is related to b then
b is related to a because if A is related to B 4 divides A-B but then 4 divides B-A
also so B is related to A.
If A divides B and B divides C that means 4 divides A-B and 4 divides B-C but if 4
divides A-B, B-C, 4 divides their sum, which is A-C, so this implies 4 divide A-C,
so A is related to C, so we have checked all the three properties. So this is an
equivalence relation. Now let us check that this is an equivalence relation, what
are we saying, let H be subgroup of G, two things are equivalent if there if this
element constructed from them is in the group H.
This is an equivalence relation. Why? First of all, is A related to A? yes, because
you have to ask yourself whether A inverse A is in H. It is in H because that is just
E, okay so that is okay. If A is related to B, that means A inverse B is in H right, by
definition. But because H is a subgroup, A inverse B whole inverse is in H,
anything in H contains its inverse, H contains its inverse also, so A inverse B is in
H means A inverse B whole inverse is in H, what is A inverse B whole inverse?
That is, B inverse times A, right? We have seen this before, A inverse times B
whole inverse is B inverse times A inverse inverse, but that is nothing but B
inverse A. So if B inverse is in A, this is the meaning of B being related to A.
Remember A is related to B if A inverse times B is in H. B inverse times A is in H
means that B is related to A. So the second property holds.
Now let’s take A related to B, B is related to C, that means A -1B is in H and B -1 C
is in H, right? A -1 B is in H, B-1 C is in H. This means because H is a subgroup it is
closed under multiplication, so if A -1 B is in H, B-1 C is in H their product is an H,
so A -1 B times B-1 C is in H, but that means is A -1 C is in H because that is the
product, that means A is related to C.
So this is an important equivalence relation on a, this is an equivalence relation and
this is some thing we will explore in great detail later. So I looked at three
equivalence relations on sets, so now some important properties of equivalence
relations. Equivalence relations
(Refer Slide Time: 12:27)
give “equivalence classes” so what do I mean? Let us say S is an arbitrary set, I am
going back to the setting of general setting of a set and an equivalence relation on
that set. So now for A in S, define the equivalence class of A, the equivalence class
of A, as we denote by this symbol [A]. So let us see the equivalence class of A is

103
denoted by A in square brackets, this is all elements in S such that A is related to B
okay. This is very simple, the equivalence class of A is the collection of elements
that are equal to B, so in our examples, what is equivalence in the first example, in
the first example where equality is the equivalence relation equivalence class is
very simple.
(Refer Slide Time: 14:06)
Right in the first example equivalence class A is equivalent to B if A = B in the
case equivalence class is all elements which are equivalent to B, equivalent to A
but this means, so this is B in Z such that B is equivalent to A, but this is same has
B in Z, B=A. so this is just nothing but A, right in this case the equivalence class is
just the set {A}.
(Refer Slide Time: 14:42)

In the second example we have defined A here S is Z, A is equivalent to B if 4

divides A-B, okay. So I am going to first write down the equivalence class of a
specific element, so that you see what is going on, so let us say, 5 that we have
seen earlier, what is the equivalence class of 5. So these are all integers such that 5
~ b, this is same as writing B ~ 5, this equivalence relation has that property right.
So this is all integers such that they are equivalent to 5, so this all integers such that
4 divides 5-B.
So if you write these down what would they be? I am just list some of them, so 0 is
not an element in this because 5-0 is 5, 4 does not divide it. So 1 is there, 5-1 is
divisible by 4, so 1 is an element of equivalence class, the previous element would
be -3, right? Because -3 is related to 5 because 5-(-3) is 5+3 = 8, it is divisible by 4,
similarly next will be 5, 9, 13, 17, 21, 25, 29, you see the pattern here, right. These
are all, the previous element will be -7.
You can see that they all integers, starting with 1 they you keeping increasing by 4,
keep decreasing by 4. We have used the notation 4Z remember 4Z, it was 4Z was
all integers which are multiples of 4. So this is –8,-4, 0,4,8 and so on right. This is
similar to that, it is not quite that, of course it is not that, but it looks like 0 is
replaced by 1, 4 is replaced by 5, 8 is replaced by 9, -4 is replaced by -3, so I can
write this as 4Z+1.
So you take 4Z and add 1 to each element, so add 1 to each element, so this is 4
Z+1 so the equivalence class of 5 is 4Z+1. We will come back to this and discuss
this in more detail, but thing to keep in mind now is 4Z the equivalence class of 5

104
the equivalence class of 5 is 4 Z+1. I will let you do this as an exercise. what is the
equivalence class of 6? is 4Z+2, for 7 it will be 4Z+3, so what about for 8? You
will see that it is exactly equal to 4Z. So and for 9 it will be 4Z+1 again, and 10 it
will be 4Z+2, so they keep coming back right, it is periodic in that sense, so as I
said, I am going to come back to this example and study it more detail later.
(Refer Slide Time: 19:09)

In the third example, the set was a group G, fix a subgroup H and the relation was
A is related to B if, so recall that in this example we have declared that A ~ B if A -1
B is in H, right, that was the example. You saw that A -1 B in H then A is related to
B that is an equivalence relation. What are the equivalence classes of? So now let
A be in G, what is the equivalence class of this. So this is all elements B in G such
that A is related to B, by definition this the equivalence relation, equivalence class,
right.
But this is all B in G such that A inverse B is in H right, this means all B in G such
that A inverse B is equal to some h, it is an element of capital H, so I am going to
give it a name, small h, so this is same as all B in G such that B is Ah, where h is in
H, right. So it is all elements in G which can be written as Ah some suitable small
h in H, this is nothing but then I can eliminate B from this whole description, it is
Ah, where h is in H, so is this clear all the equalities here.
(Refer Slide Time: 21:26)

Originally these are all elements B equivalent to A, that means A-1 B is in H, that
means A-1 B is equal to some small h, which is equivalent to if A-1 B= h then b=ah.
So I am going to rewrite this as b equals ah, but then I eliminate b from the
description completely. So this all ah, where h belongs to H, there is another
notation for this, so this is the definition, so define aH to be, this is very similar to
the way that we defined subgroups of Z. So I take this subset obtained by
multiplying every element of capital H by small a so the equivalence class of A is
in this case is just aH, so now I want to define a proposition,
(Refer Slide Time: 22:25)

105
which is, prove a proposition which is very important for us, let S be a set and let
this be an equivalence relation on S. Then the equivalence classes of elements of S
partition S. So partition S, that is, S is a disjoint union of equivalence classes.
Okay, what happens is, I will prove this in a minute, but what happens is that you
start with a set S, you have a equivalence relation on this set, then if you take
equivalence classes, they may coincide, but there will not be intersections for two
distinct cosets, so these are all equivalence classes, okay, so they partition the set,
so they partition the set meaning every element in the set is in one of the
equivalence classes. And there is no common element between two distinct
equivalence classes, so this is the proposition so let us prove this.
(Refer Slide Time: 24:29)
So first of all give if A belongs to S then A belongs to [A], right, this is because
remember one of the first property, the first property of an equivalence relation is
that an element is related to itself, so this equivalence class remember, equivalence
class of an element is all elements in the set such that B is related to A. So A is
related to A, so A is in the equivalence class of A, so certainly S is a union of
equivalence classes, correct, if you take all equivalence classes and you take their
union you get S. Because every element of S is in an equivalence class, certainly
equivalence classes are all subsets of S, so this is fine.
Now on the other hand, I claim that if equivalence class of A and the equivalence
class of another element of B are not disjoint, remember disjoint means they have
no common elements. If there is not disjoint then they are equal, so what I am
saying that if you take the equivalence class of A and equivalence class of B, if
they have something in common they must be equal.

So in fact they must be equal, so they can’t have something in common and be
distinct. What is the proof of this, so let, remember these are not disjoint, meaning
they have some common element, so let us say C is in the intersection of, see all of
this is happening within S. So this and this are subsets of S and they have
something in common, so C is in both of the equivalence classes. That means A is
related to C and B is related to C right.
Because C is in [A] , A is related to C, C is in [B], so B is related to C, this implies
that, A is related to B, because A is related to C, C is related to B, so by the 3rd
property of an equivalence relation, A is related to B, that means, A is in [B], right,
so and B is in [A], so A is in the equivalence class of B, and B is in the equivalence
class of A, so that means, because if any element, if some other element D is
equivalent to A, then it is equivalent to B also, so it is in [B], if it is equivalent to A

106
and B also, so this is fine.
So this proves that, if two equivalence classes are not disjoint, then they are exactly
identical, so this proves the claim.
(Refer Slide Time: 28:06)

Now it follows that, S is a disjoint union of equivalence classes, because, we have

either two equivalence classes are disjoint, or they are identical, so because of what
we have shown, if they have some elements in common, then they are identical,
that is what we shown here, right, if [A] and [B] are not disjoint, they have
something in common, they are identical, so if you take distinct, now take distinct,
so not equal, equivalence classes, they are all mutually disjoint, because they are
distinct, they are not equal.
They are distinct, hence any two of them being distinct are disjoint, and their union
is, because every element remember of S, is in an equivalence class, so take that
equivalence class, and that is one of the equivalence classes, that we took earlier,
so the distinct equivalence classes partition S, this proves the proposition, no need
to say really distinct, because we need not repeatedly take it, right, so these are
equivalence classes partition S is the statement, and I have proved it, and I will
work this out, in the 1st two examples, and leave the 3rd example for the next video.
(Refer Slide Time: 30:58)

So in the 1st example, remember where S was S and a ~ b, if a=b, what are the
equivalence classes here, equivalence classes are just singleton sets a, equivalence
classes of a is just {a}, so Z is a union of, so certainly it is a union of singleton sets.
In example 2, S was again Z, and a ~b, if 4 divides a-b. Here there are exactly 4
equivalence classes, so namely like the equivalence class of 0 which is 4Z, the
equivalence class of 1 which is 4Z+1, equivalence class of 2 which is 4Z+2,
equivalence class of 3 which is 4Z+3, okay.
And if you go back and see the examples, the equivalence class of 5 was actually
just 4Z+1, which is same as the equivalence class of 1, equivalence class of 6 is
4Z+2, equivalence class of 7 which is 4Z+3, equivalence class of 8 which is 4Z, so
you get nothing new, by considering other integers, so Z is the disjoint union of
4Z, 4Z+1 disjoint union 4Z+2, 4Z+3, it is clear that any integer is in one of them.
Because all you need to check is, divide by 4 and see what the remainder is, if the
remainder is 0, it is here, if the remainder is 1, it is here, if the remainder is 2, it is
here, if the remainder is 3, it is here, and they certainly have nothing in common,

107
4Z+1 and 4Z+2, cannot have any element in common, so these are the different
partitions for the equivalence class. I am going to stop the video here, we have still
to discuss the equivalence classes of the relation, given by a subgroup of a group,
that we will do in the next video, but to recall, what we have done in this video, we
have talked about equivalence classes on arbitrary sets, we looked at some
examples, and I proved a proposition saying that equivalence classes of an
equivalence relation of a set, partition that set, and which is very important for us
in the context of group theory, because it will allow us to define cosets of a
subgroup and talk about quotient groups later on. So I will stop the video here, and
continue with cosets next video, thank you.
Online Editing and Post Production
Karthik
Ravichandran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh Kumar
Web Studio Team
Anitha
Bharathi
Clifford
Deepthi
Dhivya
Divya
Gayathri
Gokulsekar
Halid
Heamvathy
Jagadeeshwaran
Jayanthi
Kamala
Lakshmipriya
Libin
Madhu
Maria Neeta
Mohana
Mohana Sundari
Muralikrishnan

108
 Nivetha
Parkavi
Poonkuzhale
Poornika
Premkumar
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirley
Subash
Suriyaprakash
Vinothini
Executive producer
Kannan Krishnamurty
NPTEL Co-ordinators
Prof.Andrew Thangaraj
Prof. Parthap Haridoss
IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resource Development
Government of India
HYPERLINK "http://www.nptel.ac.in" www.nptel.ac.in
Copyright Reserved

109
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 03
Lecture 15 – “Problems 4”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

In the previous videos we studied homomorphism of groups.

And we looked at when groups are isomorphic. So I am going to
do some exercises in this video to help understand these
concepts and so that you get comfortable working with
homomorphisms. So as a problem consider the following.
(Refer Slide Time: 00:36)
So the problem asks to describe all group homomorphisms from
the group of integers to itself. So, in other words, we are
interested in describing group homomorphism from the group of
integers under addition remember always, when I write Z, it is
always under addition, to itself. So how do you do this? So, let’s
look at what can possibly be a group homomorphism. So,
suppose that phi is a group homomorphism from the integers to
integers. And if you recall from the properties of
homomorphisms that we have already studied, we know for
example that the identify element under a group homomorphism
must map to the identity element. In the case identity element is
0. So, this is guaranteed, this is forced by a group
homomorphism. But now suppose that φ (1) is some integer. So,
φ (1) remember is some integer, so I am going to call it A.
Okay, then what is φ(n) for another integer n. So group
homomorphism properties will tell us that Φ (1+1) is equal to φ

110
(1) + φ (1), this the definition of group homomorphism. And
what is φ (1+1) that is just φ (2) because 1+1 is 2. On the other
hand, I have declared φ (1) is a, so, this is a + a, what is a + a, a
is an integer. So this is just 2a. So φ (2) is 2a. What is φ (3), 3a.
The same reasoning, so in general φ (n) is na, if n is positive
right, because φ (1) is a, φ (2) is 2a, φ (3) is 3a, φ (4) is a + a + a
+ a 4a and so on.
(Refer Slide Time: 03:20)

Now what about negative integers? Qhat is φ (-1), again another

property of group homomorphism that you recall from the
previous video. Φ (-1) must be – φ (1). I am writing in this in the
addition notation here. What we learned earlier. What we wrote
earlier was φ (a inverse) = φ (a) inverse, right that means the
symbol is less important here. The concept is image of the
inverse of a, a inverse this is the symbol of inverse of a, image
of the inverse of a is the inverse of the image a. This in this case
inverse of 1 is minus 1. So minus 1 map to inverse of the image
on 1 so, which was a. So φ (-1) is –a, so, that means φ (-2) by
the same logic as before, is minus 2a. So, you have φ (-n) is – na
so for all integers now.
Not just positive integers. For all integers n we have φ (n) is na
right, so let’s take a minute to think about what this means. I
have only prescribed what is φ(1), right and I called it a, it is
some integer I called it a. But then I have no choice on what the
other images should be, once I determine what φ(1) is, I am
forced to have φ(n)=na. How do you express this phenomenon,
so what we can write is, φ is determined by φ(1), which is to say
that once you fix φ(1), or once you give it a name, a, every other

111
image of φ is determined. Because once φ (1) is a, φ (n) is na.
So φ is determined by φ (1). All you need to specify in other
words all you need to specify to give a group homomorphism
form Z to Z, we only need to define what should be the image of
1. Right, because if I define what image of 1 is, a
homomorphism will have to be fixed once you determine what
phi of 1 is. Because once you defined φ (1), φ (n) is
automatically defined. This is the power of a group
homomorphism. Once you define what phi of 1 is, all other
images, all the other infinity integers, images are determined.
Now the next question is, what are the possible choice is for φ
(1)? So first point is this. That to give a homomorphism from Z
to Z, we only need to define φ (1). We do not sit down and see
what is φ(1) and what is φ(2) what is φ (3) what is φ(-1) we do
no need to do that. We only need to say what φ (1) is, others will
follow automatically from the definition of a group
homomorphism. So, now all we need to do is define φ (1) but
what can be φ (1) ?
(Refer Slide Time: 07:11)

Can φ (1) be any integer? It can be any integer. Because, in the

calculation that we discussed earlier a was an integer. I can put a
to be any integer. I have a valid homomorphism after I define φ
(1) to be a. So for any integer a, define a group homomorphism
φ from Z to Z by setting φ (1) = a. Once you set φ (1) =a as I
have already noted every other φ (n) is determined and give me
any integer a, I can put φ (1) =a, right, so then φ is
homomorphism. This is what I am claiming. This exercise that
we have done, no matter what a is I can declare φ (1) to be a and

112
once φ (1) =a, φ (n) will automatically be na.
Right so, for any n, so this is the homomorphism that is
determined by a. To answer the original question, remember
what was the original question, describe all group
homomorphism from Z to Z, and we have done that. So what are
the group homomorphism from Z to Z? The group
homomorphisms are determined by the image of 1. And image
of 1 can be any integer, so this is the answer to the question. The
group homomorphisms from Z to Z are determined by image of
1 and image of 1 can be any integer. So, you have group
homomorphism Z to Z are determined by integers again. So,
they are determined by Z again.
(Refer Slide Time; 10; 03)

That is the solution to this problem.

So, now I will ask you a question to continue this problem.
Which of these homomorphisms are isomorphisms? Recall that
isomorphisms are group homomorphisms that are bijective set
functions. Okay, so which of these homomorphisms are
bijections is what I am asking now. So I am going to use the
letter a as a subscript to denote that φ (a) sense want to a. so φ(a)
is the homomorphism from Z to Z sending 1 to a. And the
previous problem we have solved by saying that all group
homomorphisms from Z to itself are of the form of φ(a). So,
which of these are bijections? So for example if you take a to be
2.
What is φ (2) from Z to Z. It sends 1 to 2 and so on right, it
sends just I will write it for clarity for you. Where does it send 2,
it will send it to 4, because φ (2) must be φ (1+1). And so on.

113
And φ (n) in general so I will write that here. Φ (n) in general is
simply 2n. Is this a bijection? Think about for a minute. Recall
that bijection are 1-1 and onto. Bijections are injective and
surjective. Those were the words I used I introduced in a
previous video. Is it 1-1. It is 1-1 but not onto. That is because
(Refer Slide Time: 12:28)

it is not onto because 1 is not in the image of, clearly right,

because φ (2) sends an integer n to twice the integer n, 2n. Is
there any integer n such that 2n equals 1? There is not because
no integer multiplied by 2 is 1. So, 1 is not in the image. So, φ
(2) is not onto. So, similarly, φ (a) is not onto for a greater than
or equal to 2, right, because for the same reason φ (a) sends n. to
an, and a is greater than equal to 2 an cannot equal to 1 for any
n, right, so it sends an integer to a times this. If a is at least two,
it cannot equal 1.
So, it is not onto. What about a less than or equal to minus 2, for
the same reason, it is not onto, for a less than equal to minus 2.
Right, because again minus 2a can never be equal to minus 2n
can never be equal to 1, minus 3n can never be equal to 1. On
the other end φ (1) is simply the identity map. Right, because φ
(1) of n is n. So, it is bijecitve. And it is an isomorphism. What
about φ (-1)? φ(-1) sends n to minus n, and this already we know
is a homomorphism and you can clearly check that this is also an
isomorphism.
Because every integer is in the image, it is onto. Because, is 10
in the image? yes because -10 maps to 10. Is 100 in the image?
yes -100 maps to 100. Is it 1-1? It is certainly 1-1. All φ (a)s are
1-1, okay, that we have discussed. Kernel of Φ (a) is 0, so it is

114
an isomorphism. What about φ(0)? So we have addressed all
integers greater than or equal to 2 or less than or equal to minus
2 and 1 and -1. What about φ(0)? Φ0 (n) is just 0. So, φ0 is the
zero map. Which is to say, φ0 sends everything to, so, that is
every integer to 0.
(Refer Slide Time: 15:51)

So, certainly φ0 is not an isomorphism right, because it is not

onto and it is not 1-1. So, we have determined that only 2 of
these φ (a) are isomorphisms, Φ (1) and φ (-1). So let me
summarize the whole problem as follows. Every group
homomorphism from Z to Z is one of the, this is the answer to
the original problem. Determine all group homomorphisms from
Z to Z. They are the form φ (a) as a varies in Z. These all are the
homomorphisms. So φ (a) as a belongs to Z is the set of all
homomorphisms.
Φ (a) is an isomorphism if and only if a = 1 or a = -1. This is the
second problem that we did. It is an isomorphism if only if a is 1
or -1. Φ (a) is injective if only if a is not 0. This I didn’t
explicitly say. But this is an exercise for you. It is, the reason
that φ2 was not an isomorphism was that it is not onto, it is 1-1,
because what does φ (2) do? Φ (2) sends n to to 2n. If 2n is 0,
we have already discussed in an earlier video that a group
homomorphism is injective if and only if the kernel is 0. The
trivial subgroup, if 2n is 0, then m must be 0. So, kernel is 0.
(Refer Slide Time: 17:59)

115
And φ (a) is surjective, when is it surjective? It is surjective if
only if a=1 or a=-1. Remember that for any other integer, it is
not surjective. So, the summary of this problem which you
should carefully understand, this is an important example of
group homomorphisms and the properties that we have learned
about group homomorphisms and this description is hopefully
giving you a clearer understanding of all group homomorphisms
from Z to Z.
(Refer Slide Time: 18:40)

Okay, so, let me do one more problem, and I will do the solution
also. So that you understand how to apply the notions of group
homomorphisms that we learnt. So let G be a group, so this the
second problem of this video. Let G be a group and let a be an
element of G. Let G be a group, let a be an element of G.
Suppose that order of a is r. So, what is order of a? I am going to
recall it here. Order of a is the smallest positive integer n let’s
say d such that a power d is the identity element. This was our
original definition of order. It is the smallest positive integer
such that a power that is the identity element. Here I am saying
that order of a is some integer r. So, obviously it is a positive
integer. In other words, I am saying that some power of a is
identity, some positive power and now suppose that a power n is
also identity, for some other positive integer n. Then show that r
divides n, so this is the problem. Okay, so the problem is saying
that if order is given to be a number r. and a power n is identity
for some other number n, all positive numbers, then r must
divide n.
Okay, what is the solution of this? I am going to recall for you,

116
another description of the order that I gave after I defined
homeomorphisms. Consider the homomorphism from Z to G
which sends some integer m to a power m. In an earlier video,
already we saw that φ is a group homomorphism. That we
already saw in a previous video. We also saw that if you
consider the kernel, so kernel of φ is rZ, where r is the smallest
positive integer such that a power r is identity. Remember in
general there is another possibility kernel of φ may be either rZ,
or kernel of φ is 0. This is is also possible in general because
there may not be any kernel of this map. That means a power m
may never be identity for any positive integer. Okay, so this is a
possibility in general. But in our problem kernel φ is not {0}. I
should really put as a set.
Because kernel φ is 0 is a possibility general. But in our problem
it is not possible. Because in our problem remember we are
given that order of a is r, where r is a positive integer. Order has
to be either 0 or positive integer always. And in this problem we
are given that order is a positive integer. So kernel φ is not 0
because r belongs to the kernel. So, note that r is a positive
integer. So, kernel is non-zero. That does not happen. So kernel
is this is not going to be a case in our problem. Kernel is rZ. So
in general, kernel is in generated by the order of a, okay. Now
let us look at the problem more carefully. So r is order of a, and
a power n equal to identity for some integer n. Since a power n
is identity we have that n belongs to kernel φ. Because what is
φ? Φ sends to n to a power n. So φ (n) is a power n which is
identity, so n belongs to kernel φ. But kernel φ as we discussed
earlier is rZ. So, n belongs to rZ. So, n belongs to rZ. What does
it mean? Then n=r times m, for some m, right, because
remember recall what is rZ? rZ is the set of multiples of r. It is r,
2r, 3r, –r, -2r, -3r, and so on. If n belongs them then n is the

117
form.
(Refer Slide Time: 25:34)

So n is the multiple of r. So, r divide n, which is exactly what the

problem asked us to do. It asked us to show that r divides n,
which is what we have showed. Okay, so keep in mind that, if an
element has order r and a power n is identity the r is divides n.
Basically what happens is that if you take a and you write
multiples of a, not multiples, sorry, powers of a. So you have a
power 0, which is identity then you have a, a2, a3. There are two
possibilities either you will never 1) we never encounter e in this
sequence right, so it keeps going, a power 100 and a power1000
so on. So, there are all distinct and e is never there. Clearly the
group must be infinite in this for this case to occur. The other
possibility is 2) e occurs, I mean of course when I say e never
occurs e certainly occurs at the beginning. We never encounter e
in this sequence after a power 0, we do have e in this, but if this
sequence actually starts with a, not with a power 0. In this
sequence either e appears or e does not appear, if e does not
appear they are all distinct. And the group in particular must be
infinite. Or e occurs in this sequence and order of a is the first
place where e occurs. Again I am not considering this. Let me
not even put that in the sequence. I keep looking at a, a2, a3, a4,
a5 and so on. And the first time I see e, if it is a power r, then r is
order of a. And the point is the sequence then repeats. So
suppose a power 5 is identity. So the sequence is a, a2, a3, a4,
a5, a, okay, which is e as I said I am assuming this. So example
and then what is a power 6? this is just a power 5 times a. So,
this is just a. So, the 6th element again a. Then a2, a3 this is same

118
as a7, this is same as a8, a4 is a9, a5 a10 which is also again
equal to a. So, the sequence is just
(Refer Slide Time: 29:16)

repetitions of this, so this block keeps repeating. This segment

keeps repeating forever. So what we are saying is if a power100,
a power some number is e, that number must be a multiple of 5.
So this is e. In this block e must appear at the last spot in each
block. And that spot is the multiples of 5. So this is the 5 th
position, 5th position, this is the 10th position, this is the 15th
position. This is the 20th position.
Okay, so the problem is essentially encapsulating the detailed
data that we have just discussed. If a power n is e, then it must
be a multiple of the first time uncounted e in the sequence. And
let me quickly illustrate both of these possibilities, both of these
possibilities occur, for example if you take G to be Z and a to be
1. Or let’s say 3. If you take 3, if this sequence again we
remember in the general case we use the multiplicative notation
a, a2 a3. But when we are dealing with integers we use additive
notation. So the sequence will be 3, 6 twice, and two times and
9, 12, 15, 18 and so on. So, we never encounter 0.
This is the possibilities number 1 here. So this is a sequence of
distinct integers never containing 0. So this is the first
possibility. On the other hand, if you take G to be is some other
group. For example, if you take G to be the group of 4th roots of
unity, 4th roots of unity, that we encountered before, and you
take a to be I. So, you start with I, then you take I2 which is
minus 1. But once you come to I power 4 it is 1. So, this is the
segment that keeps repeating. So the next I to the 5th will be I

119
again. I to the 6th will be minus1. I to the 7 th will be minus -I.
and I to the 8th will be 1 again.
(Refer Slide Time: 32:10)

So this will be repeated. So the segment that keeps repeating is I,

minus 1, minus I and 1. So, then you repeat this. And then you
repeat this. You repeat this forever and in particular we know
that order of I is 4 because 1 is appearing in the 4 th spot, 8th spot,
12th spot, 16th spot and so on. So, this illustrates for you the
point of this exercise. Okay so let me now end this problem by
doing a related problem,
(Refer Slide Time: 32:43)

which exactly uses this and this also makes use of the notion of
group homomorphism. So let’s say G to G prime is a group of
homomorphism, let us say, I take an element in G, suppose that
the order of a is r, so you could have simply written is as order
of a is r. The problem is, what can you say about order of φ (a)?
So, the problem is if you have a group homomorphism and an
element is fixed which has some finite order, what is the order
of the image? Can you say anything about the order of the
image. Order of a is r, what is the order of φ (a). How is it
related to r in other words, that is the question.
(Refer Slide Time: 33:49)

120
So, solution is the following. Let’s recall that order of a is equal
to r implies, a power r is identity. Now let me, this is the identity
of the group G, that is because a is inside G. Now let us apply φ
to both sides. Right, so φ a power r is equal to phi of eg. Now
continuing here and using the properties of group
homomorphism, φ (a power r) is just phi of a whole power r.
And what is φ (eg), that is eg prime right, because the identity
element under a group homomorphism maps to identity element.
So φ (a) whole power r is eg prime. By the previous problem
which I have written here, what is the previous problem in any
group if a power n is e, then order of a divides n. So here φ (a)
power r is eG prime. So, order of a order of φ (a), whatever it is,
divides r. That is all we can say. So, what can you say about
order of φ (a)? We can say that order of φ(a) divides r.
So this gives you a restriction on the possible images of
elements of a group under a group homomorphism. If an
element has order 5, its image has to have an order that divides
5. So it has to be either 1 or 5. So in particular, if φ is a group
homomorphism, let me shorten it like this, group
homomorphism and a is an element of G has order 5 then φ (a)
has order 1 or 5. Right, it is an immediate consequence of the
problem, because whatever is the order of φ (a), must divide 5.
5 being a prime number, it is either 1 or 5. So and as a
supplementary exercise, this is a trivial exercise that I have
repeatedly used. Let G be any group and let a be an element of
G.
Then order of a is one if and only if a = e, okay, so this is easy
for you to do. Only element which has order 1 is e, okay,
coming back to this example φ (a) is order 1 or order 5, so in
other word φ (a) is either identity element or its order must be 5.
So, this is very useful to keep in mind, order of any element

121
must divide the order of the element. Order of phi of a must
divide the order of a.

So, I just recall for you, what is order of r is a. So order of φ (a)

divides order of a, okay. So let me stop the video here. And
hopefully this gave you an idea of how to work with group
homomorphisms and apply various properties that we have
learned, so in the next video I will continue with my study of
cosets that I have introduced, study of equivalence classes that I
have introduced in a previous video. Thank you.

122
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 03
Lecture 16- “Cosets and Lagrange’s theorem”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE
So, I am going to continue now my discussion about the
equivalence classes and how this leads to the notion of a
quotient group. So, recall from video on equivalence classes, an
equivalence relation on a set is a relation
(Refer Slide Time: 00:31)

where you can say, if S is a set, we can say that two elements are
equivalent and in order to be an equivalence relation it must
satisfy that any element must be related to itself. If an element is
related to A is related to B then B is related to A, if A is related
to B, B is related to C then A is related to C. So, these are the
properties of an equivalence relation. We discussed some
examples of this and the most important example in our situation
would be the relation defined by a sub-group of a group.
So, that I will recall in a minute, but an equivalence relation, the
main fact that I want you to recall from that video and that we
will use today is an equivalence relation, an equivalence relation
partitions. So, an equivalence relation on a set S partitions S. So,
I discussed this in all those examples of equivalence relations
but if S is given like this, you consider equivalence classes.
What are equivalence classes?
So, equivalence classes are simply, if you fix for A in S

123
equivalence class is all elements in S which are equivalent to B,
whether I write A is related to B or B is related to A is not
important because of the reflexive property of or the symmetric
property of the equivalence relation. So, these are the
equivalence relations, classes. So, you have A1, A2 and A3 and
so on. And the important property is either if you are given to
equivalence classes, they are, given two equivalence classes,
let’s say the class of A and the class of B, then we have two
possibilities, then 1) either the class is equal to itself each other
the class A is equal to the class B ([A]=[B]).
So they are identical or 2) they are disjoint. They have nothing
in common. So, this is a very strong condition, right? You have
either disjoint sets or they are identical. So, if you take distinct
classes, if A1 class is different from A2 they cannot have
anything in common. So, you could cover the entire set like this.
So, this is how a set partitions,
(Refer Slide Time: 03:39)

an equivalence class on a set partitions that set. So, the most

important example for us and all the examples that we looked at
are examples really of this example with different groups and
different subgroups. Fix a group, so fix a group G and a
subgroup H of G. Fix a group G and fix a sub-group H of G. So,
we say that I think I defined if A and B are in G then A is related
to B if A−1B is in H.
Right, this is the equivalence relation that we checked, is an
equivalence relation. So, what are the equivalence classes? And
we discussed this, I think. Equivalence class of A is all B in G
such that A −1 B is in H. But this is same as AH because if A −1 B
is in H, A −1 B is equal to small h, so B is small a times H. So, B

124
is related to A if and only if B is A times an element of H. So,
this is just this set.
This set remember is my notation, the definition is this, this is
the definition. So the equivalence class of A is simply (aH). And
the most important definition now I want to give you is the
following. If H is a sub-group of a group G, the “left cosets of
H” are subsets, so, left cosets are simply, you take an arbitrary
element of the group and multiply by (aH). So, left cosets are
these.
(Refer Slide Time: 06:21)

Similarly, you can define “right cosets” and not surprisingly,

these are subsets obtained by multiplying by A on the right, Ha
are right cosets. So left cosets are aH right cosets are Ha. So
what we have done is for the equivalence relation of this
example, for the equivalence relation above by which I mean the
previous page, A is related to B if A −1 B is in H, for that
equivalence relation, the equivalence classes are simply left
cosets.
Correct? So equivalence classes of this relation are the left
cosets of H. So as I said, the left cosets and right cosets are very
important for us and we will constantly keep dealing with them.
So you should carefully think about what they are, and they are
very simple really, the left cosets are subsets of the form aH and
rights cosets are subsets of the form Ha and I will do some
examples in a later video to describe these in specific groups,
and hopefully that will become clearer to you then.

So, now coming back to the equivalence relation that we are

discussing for that equivalence relations the left cossets are the

125
equivalence classes. So, now I want to do an important theorem,
this is the first important theorem really of the course and in
order to prove that let me observe the following. So, we have
noticed two facts. We have noticed already one fact that, since
equivalence classes partition the set, in our situation the left
cosets of H in G partition G.
So left cosets of H are equivalence classes for an equivalence
relation. So, they partition H, which is to say, this means G is
the disjoint union of left cosets. The word disjoint means that if
you have two distinct left cosets, they disjoint. They have
nothing in common. Remember this is automatically true for us
because of the property of equivalence classes for an arbitrary
equivalence relation on an arbitrary set, two equivalence classes
are either equal or they are disjoint. So, in particular in our
situation we are looking at a special equivalence class on a
group and the equivalence classes are the left cosets so they
partition the group. So, the group is the disjoint union of left
cosets.
(Refer Slide Time: 09:57)

Now let me prove an important proposition. The number of

elements, so fix let H and G, by which I mean G is an arbitrary
group and H is a sub-group of G and let a be in G. Then the
proposition says the number of elements, then the number of
elements in the left coset aH is equal to order of H. Recall order
of a group is simply the number of elements. So, order of a
group is the number of elements of H.
Remember I am not saying, order of aH because aH is not a
group, aH is just a left coset. So, the number of elements of aH
is equal to the number of elements of H, that is the proposition.

126
And the proof is very simple. Consider the map from H to aH
given by a small h goes to small a times h. This is a set map.
Right, this is a map of sets. Remember aH is actually nothing
but a set. It has no further structure; it is just a subset of the
group G.
It is a left coset, it is a subset of group G. So, this is a map of
sets, give me an element h I will multiply by a and map into this.
This of course belongs to aH. I claim that this is a bijective map.
Why is this? First of all, is it onto? Yes, because what are the
elements of aH? In previous slide I wrote that right, aH or ah as
h varies. So, you give me an element of aH that is of the form a
times h. So, h maps to that. Right, so it is certainly onto, by
definition, everything in aH is a multiple of something in H.
Is it 1-1? Let us prove this. Suppose h1 goes to a h1 and h2 goes
to ah2. Let us suppose that they are equal, ah1 equals ah2. But
then the cancellation property of groups that we have learned
long time ago gives me h1 equals h2, by cancelling A. So, it is
1-1 also, so it is a bijective map. If you have two sets which,
there is a bijection between them that means they have the same
number of elements.
So, H and aH have the same number of elements. Right, because
if there is a bijection that is by definition that means they have
the same number of elements. Now let us look at, see one thing
that I should mention here, that I should have been careful when
I stated the proposition. I am going to assume that, though it is
also true in general, G is a finite group, because this is useful
only in this situation, G is a finite group.
So, G is has only finitely many elements, in particular H has
only finitely many elements because H is a sub-group of G. So,
this proves the proposition right. The proposition is proved now.
(Refer Slide Time: 14:20)

127
Next we have two facts. So, let me draw a picture here. So, this
is the group G. You have cosets, remember H itself is a coset
because H is nothing but eH. So, that is one coset. You have
some other coset a1H, some other coset and so you have several
cosets and we have two facts: the first fact is that G is the
disjoint union of left cosets. So, these cosets cover G. So, you
have this coset union, this coset union this coset union, this coset
and so on is all of G.
And the previous proposition says that any two cosets have the
same number of elements, because if you aH has the same
number of elements as H, a1H has the same number of elements
as H but so does a2H, a2H also has same number of elements as
H. This proposition works with an arbitrary element a of G. So
any two cosets have the same number of elements. So, now let
us think about what could be the number of elements of G.
So, we have, G is a disjoint union of H, union a1H, union a2H,
anH, because G is a finite group, remember that G is a finite
group. So, it has finite many cosets right because they are only
finitely many elements. This is my symbol for disjoint union.
So, it is a union a1H union a2H union anH. How many elements
are over here? Remember n, let me denote this by a2 because
this is taken to be a1, n is the number of left cosets of H in G, n
because a1H, a2H, a3H, anH, it is a number of left cosets.
So, n is the number of cosets that are distinct among all the coset
of H in G. So, now because G is partitioned by these cosets,
cardinality of G, number of elements of G, which is order of G
is the number of elements of, let me use for simplicity, the same
symbol within vertical bars, number of elements of G because G
is a disjoint union of these things. Every element of G is in

128
exactly one of them right.
So, number of elements of G is number of elements of this, plus
number of elements of a2H plus number of elements of a3H plus
number of elements of a4H and so on. So, number of elements
of G is equal to number of elements of e H, number of the
elements of a2H, number of elements of a3H and finally number
of elements of anH. But by the proposition of the previous slide,
number of elements of a2H is also the number of elements of H,
a3H is also anumber of elements of H, anH is also number of
elements of H.
It does not matter what cosets we are considering they are all H.
How many factors here are there? There are n terms. Right, so
cardinality of G is the cardinality of H added to itself n times.
So, we have that cardinality of G is equal to n times cardinality
of H, and this leads us to state our first important theorem.
(Refer Slide Time: 18:41)

Before that let me define the following number, the number of

cosets, I am going to stick to left cosets for this calculation, and
for this theorem, because everything that we have done is about
left cosets. The number of left cosets of a group H in G is called
the “index of H in G”. Index of H in G is the number of left
cosets of H in G, it is denoted by this symbol, [G: H]. So, the
previous slide we ended with this statement, cardinality of G is n
times cardinality of H, where n is the number of left cosets.
I have given this a name now. So, we have a “counting formula”
so this is an important counting formula in group theory. We are
going to encounter other counting formulas which are formulas
to count elements of a group in various contexts. The first
example of counting formula is this. Order of a group is the

129
index of the sub-group in H times order of H. So, this is the
counting formula.
This is a very important formula that we have to use repeatedly
in our future videos. So, please pay close attention to this and try
to understand this carefully. I have already proved this. The
previous work is a proof of this. Okay, if you are not clear you
should go back and listen to this again and convince yourself
that we have the counting formula, it says that cardinality of the
group or the order of the group is the index of H times the order
of H. Here H is a subgroup of G. Okay, so as an immediate
corollary of this I am going to write a very important theorem,
follows immediately.
(Refer Slide Time: 21:16)

It is called the Lagrange’s theorem. So, this is one of the first

important theorems in group theory. Again I keep repeating this,
at some point in these videos I have started assuming G is a
finite group. So, you can talk about cosets and equivalence
relations always without assuming that G is a finite group but
once I started counting number of elements, I have assumed that
G is a finite group.
Everything that follows is about finite groups. For example the
statement about counting formula is a statement for finite
groups, right because if G is not a finite group there is no sense
in the statement, because this is an infinite number and you
cannot write something. So, if G is a finite group you can make
this. So, similarly Lagrange’s theorem is a statement about finite
group. So, I will record that here.
Let G be a finite group and let H be a sub-group of G. Then the
order of H divides the order of G. Okay, so written differently,

130
|H| divides, because the symbol for the order is this, order of H
divides |G|. This is as I said is a very very important theorem in
group theory and it gives you a lot of flexibility in working with
groups. And what is the proof? Immediately follows as I said
from the counting formula because |G| is something times |H|.
We actually know what this number is but it is immaterial for
the Lagrange’s theorem. Right, this follows easily from the
counting formula. Correct, because |G| is the product of |H| and
something else. So, that means |H | divides |G|. So, I am going to
work out some simple examples to illustrate this but pause for a
minute here and then think carefully about what Lagrange’s
theorem is saying.
It says that if you have a finite group, it puts a major restriction
on possible orders of subgroups. Immediately you can conclude
that if, for example G is a group of 6, for example S3, it cannot
have a subgroup of order 4. Right, because 4 does not divide 6.
So that is what Lagrange’s theorem says. Okay, so this is our
first application of the study of equivalence relations and left
cosets. It gave us a very important result about orders of a finite
group and orders of its subgroups.
Online Editing and Post Production
Karthik
Ravichandran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh Kumar
Web Studio Team
Anitha

131
 Bharathi
Clifford
Deepthi
Dhivya
Divya
Gayathri
Gokulsekar
Halid
Heamvathy
Jagadeeshwaran
Jayanthi
Kamala
Lakshmipriya
Libin
Madhu
Maria Neeta
Mohana
Mohana Sundari
Muralikrishnan
Nivetha
Parkavi
Poonkuzhale
Poornika
Premkumar
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirley
Subash

132
 Suriyaprakash
Vinothini
Executive producer
Kannan Krishnamurty
NPTEL Co-ordinators
Prof.Andrew Thangaraj
Prof. Parthap Haridoss
IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resource Development
Government of India
HYPERLINK "http://www.nptel.ac.in" www.nptel.ac.in
Copyright Reserved

133
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 03
Lecture 17– “S_3 revisited”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE
Okay I want to illustrate this with an example, so I am going to
revisit.
(Refer Slide Time: 00:24)

one of the first groups that we studied, S3 revisited, so S3

remember is the symmetric group, the notation the terminology I
want to use is symmetric group on 3 letters, in other words, it is
the group of bijections from a 3 element set to itself, okay in my
earlier notation which was given in my first video, I concretely
told what these 6 bijections are and gave them the names f1, f2,
f3, f4, f5, f6.
Now I want to give a more logical, more clear, notation for this
group, because obviously this is not useful, right, now what is
f2, you have to go back to what I called f2. So instead of doing
that I am going to give you a different description of S3 and in
subsequent lectures I want to use that description. So I want to
introduce a new notation to describe elements, I want to use a
notation which, which is useful to understand what that element
is doing because here, I have no idea what f2 does, no idea what
f3 does. So my notation I want to, immediately looking at the
notation I need to know what element what it does as a function
and I also do not want unnecessary information.

134
So let me introduce this , this is called the cycle notation, so if
you recall what f2 was, I will just do it for 1 element or 2
elements. f2 was the function with sends 1 to 2, 2 to 1 and 3 to
3, right. So this was a function, so let us look at this carefully
and how to minimize the data that I have written all these things,
a lot of it is unnecessary, for example 3 goes 3, that I already
know.
I would like not to mention that at all in my description, so in
the new notation, let me write this as simply, (1 2), so all my
elements of S3 will denoted by elements 123 within brackets,
okay, and I will tell you how to read this. First of all, any
number that does not appear here is supposed to go to itself, so 3
is not appearing here, so this is supposed to stand for f2, 3 is not
appearing here so that means 3 goes to 3 and where does 1 go? It
goes to the next number there, it goes to 2, where does 2 go? It
does not have anything to the right, so you cycle back to the
beginning of the expression. So 2 goes to 1, so (1 2) sends 1 to 2
, 2 to 1, and 3 to 3. So you can definitely agree with me that this
is a much simpler notation than writing all this data here, and the
advantage, of course, is also that just by looking at you know
what it does as a function: it sends 1 to 2 , 2 to 1 and 3 to 3, so
you can replace f2 with (12), in this same way you can replace,
use this notation.
(Refer Slide Time: 04:24)
What does f3 do? f3 sends 1 to 3 , 3 to 1 , 2 to 2, so in this
notation, in the new notation, I am introducing, I start with 1,
okay, where does 1 go under this? It goes to 3, so I put 3 next to
1, and where does 3 go, it goes to 1, so I do not want to write a
number again, so I will remove 1 and I start the bracket, I put 1,
1 goes to 3, so I will put 3 there, but 3 goes 1, so I close the

135
bracket because then once you look at the last number on the
bracket, before the bracket it goes to the first element in that
thing, in other words, it goes to 3, so remember in this new
notation if I use (31), it will be the same element, because 3 goes
to 1 and 1 goes to 3, and I am not going to introduce, I am not
going to write anything about 2 in this notation, because 2 is
supposed to go to itself, in the previous term also, (12) also can
be written as (21), the point is 2 goes to 1 , 1 goes to 2.

So it will take time for you to get used to this, but this is a very
useful notation to represent bijections of sets. 1 goes to 3, 3 goes
to 1, 3 goes to 1, 1 goes to 3, that’s the same function. What
does f4 do? It fixes 1 and it sends 2 to 3 , 3 to 2 and again it is
clear, 1 goes to 1, so I do not want to even write 1 in the new
notation. So I start with 2, it goes to 3 , 3 goes to 2, so I close the
bracket. f5, f5 was, 1 goes to 2 , 2 goes to 3, 3 goes to 1, so this
is more interesting.
Now let’s put the bracket and start writing the cycle notation. I
put 1, 1 goes to 2, so I put 2, 2 goes to 3, now I cannot close the
bracket like earlier, because 2 does not go to 1, 2 goes to 3, so I
have 3, 3 has not appeared so far, so 3 must be put next to 2, but
3 goes to 1, so 1 is the first element, so I close the bracket. But
again just like before I can cycle them and I do not change the
function, so for example, I can do this also write this as (231): 2
goes to 3, 3 goes to 1, 1 goes to 2, I can also write this as (312),
so there is some ambiguity in which order you, what you start
with, but there is no ambiguity in the order in which they appear.
So the correct way to think of them is (123), so this is 123 if you
go clockwise 123 it does not matter where you start, if you start
with 1, you have this description, if you start with 2 you have
this description, but 231, it’s important to traverse these things

136
in the correct order, if you go clock-wise, 231 or you can started
3, 312.
So that is this cycle, here it is 2 and 3, then you traverse in
clock-wise order, so 23 or 32. Finally what is f6? f6 was 1 goes
to 3, 3 goes to 2, and 2 goes to 1, so if you write the cycle
notation, you put 1 then you put 3 because 1 goes to 3 ,3 goes to
2, and 2 goes to 1 again, so you close it. So in the cycle notation
you have 1 and 3 and 2, so (132) is same as you can start with 3,
(321) or (213). Remember that in no convention these 2
elements will be same, because 1 goes to 2, whereas 1 goes to
3, here only thing you have to keep mind is in a cycle notation,
these are called cycles, the image of an element is determined by
what comes after it. In this notation, image of 1 will be 3, image
of 3 will be 2 and if the element is at the end of the cycle and the
next element next you see the bracket closed, its image is the
first element of this cycle, so 2 goes to 1, 1 goes to 3, 3 goes to
2.
(Refer Slide Time: 08:57)

So in this notation now, S3 is, identity element I will now as

always denote by e, you have {e, 12,13,23,123,132}, right, and
you agree with me hopefully that it is much easier than before,
earlier I was calling them f1, f2, f3, f4, f5, f6 and if you just tell
you what is f2, you have no way of remembering it right, you
have to go and see where we have defined what it is f2 and see
what it is. Now it does not matter, I can forget that earlier
notation, if I consider now the function 23, it is clear 2 goes to 3,
3 goes to 2, 1 goes to 1, 123 sends 1 to 2, 2 to 3 , 3 to 1 and so
on. So let us I haven given this description because this is much
more natural to work with this and especially when we are

137
writing cosets and so on, it is useful to have this notation.
So now let us work out the cosets of a particular subgroup and
verify counting formula and the Lagrange’s theorem in this case.
So this is my ambient group G, consider a subgroup, I can take
anything I want, so let me take this sub group, as an example,
given by H= {e, 12}. So remember we have already verified
that this is a subgroup, because 12 times 12, I should also tell
you how to multiply these things. We start with the right hand
side, 12 times 12, and you trace what happens, so for example,
under the product where does 1 go? 1 goes to 2 under the first
element then 2 goes to 1, under this, so 1 goes to 1, so there is
no need to write it, if a number is not there it is basically by
itself within the bracket. 2 goes to 1 under the first function and
1 goes to 2 so 2 goes to 2 under the product. Similarly 3 goes to
3 this is e, okay, so just as a practice, for example ,we will do
this as part of the writing cosets, but what would be 23 times 12
, 23 times 12. As I said, you start on the right hand side element
and work from there. Under the product where does 1 go? You
do first where does, you ask yourself were does one go under the
first function, first element 1 goes to 2 under the first element
but 2 goes to 3 under the second element so in the product 1
goes to 3, so you have 13, in the product, now where does 3 go?
3 goes to 3 under the first element, because 3 is not appearing
in the first element so 3 goes to 3 and 3 goes to 2 in the second
element so in the product 3 goes to 2 okay.
Let me see if I can close the bracket. Where does 2 go under the
product? In the first entry, in rightmost entry element 2 goes to 1
right and where we does 1 go under the left entry element 1 goes
to itself, so 2 goes to 1 in the product, so the product of 23, 12 is
132.

138
So now let’s come back and calculate the left cosets of H in S3.
So this is the problem. Compute the left cosets is H in S3. What
are the left cosets? Remember left cosets are elements of G
multiplied by h, element is on the left hand side, so this is eH,
12H, 13H, elements are now defined like this, 23H, 123H,
132H, so these are the left cosets. But remember, you will
expect that some of them will coincide with each other, so that
what we have to figure out.
(Refer Slide Time: 13:26)

So let us compute what is 12H? So 12H remember recall in

general what is the left coset at aH, in the general case, it is the
set consisting of a times small h, in our example, 12H is and H
was {e, 12}. I will fix that here, H is {e,12}. 12H is 12 times e
and 12 times 12. I am going to multiply each element of H by
12, so 12 times e and 2 times 12. And what is 12 times e? e is
the identity element, so that is 12, and 12 times 12 as I worked
out in the previous slide, this is just e, 12 has order 2, so 12H is
just {12, e} which is exactly equal to H. So the first two cosets
here coincide, so eH is seem as 12H.

Now let us compute 13H. What is 13H? 13 times e and 13 times

12. 13 times e is 13 because e is the identity. What is 13 times
12, and I am going to use the same procedure that I described
earlier and you have to do this repeatedly if you are new to this
to get used to the product of this cycle notation.
You have to ask yourself put a bracket, put 1, you can take any
number right, so let us start with 1 and let us determine the
image of 1 under the product, so it is a composition of functions,
that is all that is happening here. 1 goes to 2 under the first

139
function, 2 goes to 2 under the second function so 1 goes to 2
under the composition of the two functions. Where does the 2
go? 2 goes to 1 under the first function and 1 goes to 3 under
the second function so 2 goes to 3 under the composition.
So I put 3 there. 3 goes to 3 under the first function because 3 is
not appearing in the cycle notation and 3 goes to 1 under second
function, so you close. So 13,12 is 123. So this is a coset, so
this is what 13H is. Now let us compute 123H. 123H is 123
times e and 123 times 12. Before I compute this, let me just
write 123 times e is 123. Before I compute this product 123
times 12, let us look at the previous coset 13 and 123, and this
coset has 123.
So the coset 13 times H and the coset 123H have something in
common, 123 is common to both of them, and remember our
general theorem about equivalence classes of equivalence
relation which applies to left cosets of a subgroup in a group: if
2 cosets have anything in common they must be identical, this
coset has these 2 cosets have 123 in common, so you expect
from that result that these are identical.
So you expect in this place 13. It must be 13, if it is anything
else then these cosets have something in common yet there not
equal. it has to be 13, but let us verify just for surety that it must
be 13. So in other words, what is 123 times 12? Following the
same recipe for the product, I put 1 here 1 goes to 2 under the
first function 2 goes to 3 under the second function so 1 goes to
3 under the product. 3 goes to 3 here and 3 goes to 1 here so 3
goes to 1 under the product, so we finish the bracket. Now what
happen to do 2 goes to 1 here 1 goes to 2 is fixed and no need to
write that and sure enough, this is 13. So these 2 cosets are same
so this is same.
Similarly we have to compute, let’s say 23 times H, what are

140
these? What is this? This 23 times e and 23 times 12, because H
is 12, and 23 times e is 23 and 23 times 12, I will let you
calculate using the same procedure it will give you 132, okay.
Now the only remaining coset we have to compute is, remember
there are 6 potential cosets, eH,12H,13H, 23H, 123H, 132H, we
have identified the first two, so let I will write this later, we have
identified the first two and 13 happen to equal 123, this and this,
now let us to 23 and 132, what is 132H? This is 132 times e and
132 times 12, again remember H is {e, 12}. So 132 times e is
132, and exactly as before because they have something in
common, they must be identical which you can also verify by
explicitly multiplying this out and I strongly urge you to do this
you on your own, 132 times 12 it is actually 23, so this 2 cosets
are equal, so there are, let us now list all the distinct cosets
(Refer Slide Time: 19:13)

distinct, okay, list all the left cosets, again I am working with
this specific example. H is {e, 12}, okay, so this is eH and 12H,
this is one coset , this is the first, okay, this is first left coset I
wrote. The second one is 13H which also happens to be same as
123H, this is the second left coset and finally we have 23H and
that is equal to 132H, this is the third left coset. Remember that
S3 has 6 elements, so potentially there are 6 left cosets right
because of each element you multiply by H, so eH, 12H,13H,
23H,123H, 132H. So there are 3, 6 potential things but because
of identifications because they are not all distinct, you have only
there are only, 3 left cosets, there are 3 left cosets of H in G, in
other words, the index of H in G is 3, in my notation that I
developed earlier, [G :H] is 3, that means G was S3, so let me
write S3 colon this particular subgroup is 3. So is the counting

141
formula verified in this case?
(Refer Slide Time: 21:16)

Recall what is the counting formula? It says order of the group

equals the index of H in G times order of H, in our case order
was, G was S3, so this is 6, index we just calculated is 3, and
what is the order of group? It is 2, there are 2 elements, so 6 = 3
times 2, this is correct, counting formula is verified and
Lagrange’s theorem is also verified here, of course, for that you
don’t need to do any coset calculations, right, because order H
which is 2 divides order of S3 which is 6.

So as an exercise, I will strongly urge you to do this, to get

comfortable with calculations of cosets, compute the left cosets
of H, new subgroup I take is, H is {e, 123,132}. So first of all
verify that this is a subgroup and compute its cosets in S3.
Counting formula tells you how many you expect, counting
formula says how many cosets should be there? What is order of
H order of H, 3 elements are in H, so order of H is 3, order of S3
is 6. The counting formula says 6 equals index of H in G times
cardinality of H, in this case cardinality of, order of H is 3, so
the number of cosets should be 2. So verify this, by explicitly
computing the left cosets, see that there are 2 left cosets of H in
G. So this tells you that, there are 2 left cosets of H, this
particular H in S3, so this is an exercise for you I will let you do
this I am not going to discuss this, exactly the same calculation,
but in this example I also want illustrate the fact that left cosets
partition of your group.
And that is clear right, because S3 remember is {e,12,13
23,123,132} and left cosets cover all the 6 elements, so left

142
cosets are in the previous slide we computed this, eH and 12H
are same time so that, and what is eH? eH was, eH was {e, 12},
so 2 elements are in the first left coset, there are two more
elements 13 and 12 are in the second left coset, 23 and 132 are
in the third left coset, so we also see, so maybe I will just record
this here.
(Refer Slide Time: 24:46)

In this example that we worked out, we also see that each left
coset, this example is not the exercise, the example of, in the
example H was {e, 12}, each left coset has 2 elements,
remember this is what you would expect because of a theorem
that I proved earlier, because H has 2 elements, each left coset of
H has 2 elements. I want to each left coset of H has 2 elements,
which is what we noticed in the calculation.
Every left coset, 12H is 12 and e, 13H is 13 and 123, there are 2
elements and there are 3 left cosets, each coset has 2 elements,
there are 6/2 = 3 left cosets, again this is counting formula right,
number of cosets is the number of elements in the group divided
by the number of elements of the subgroup, there are 3 left
cosets and left cosets partition, this also we saw, so all 3
properties we proved in general, we have illustrated them in this
example and I want you to do the exercise with a different
subgroup of S3, namely a subgroup of 3 elements. Here what do
you expect? Each left coset should have 3 elements, there should
be 2 left cosets and again all the left cosets should partition the
group, so do the exercise, compute everything, all the left cosets,
and verify that all the properties are satisfied.
So in this video we have done a very important theorem called
the Lagrange’s theorem which says that a group, if you have a

143
finite group G and a subgroup of H of G, the order of H divides
the order of G. This is an extremely important theorem and we
proved that as a consequence of counting formula, which is also
important, which says that order of a finite group is the product
of order of the subgroup H and the index of H in G, and we have
illustrated all these results in the case of S3 and in order to do
that, we have found a new description of elements of S3 using
cycle notation, which is very important and that we use
repeatedly in future and worked out all the left cosets of a
particular subgroup H of S3, thank you.
Online Editing and Post Production
Karthick
Ravichandaran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh kumar
Web Studio Team
Anitha
Bharathi
Catherine
Clifford
Deepthi
Dhivya
Divya
Gayathri
Gokulsekar
Halid
Hemavathy

144
 Jagadeeshwaran
Jayanthi
Kamala
Lakshimipriya
Libin
Madhu
Maria Neeta
Mohana
Mohana Sundari
Muralikrishnan
Nivetha
Parkavi
Poonkuzhale
Poornika
Premkumar
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirley
Sorna
Subash
Suriyaprakash
Vinothini
Executive Producer
Kannan Krishnamurty
NPTEL Co-ordinates
Prof. Andrew Thangaraj
Prof. Prathap Haridoss
IIT Madras Production

145
 Funded by
Department of Higher Education
Ministry of Human Resource Development
Government of India
HYPERLINK "http://www.mptel.ac.in"www.mptel.ac.in
Copyrights Reserved

146
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 04
Lecture 18 – “Problems 5”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE
Okay so in the last few videos, we have learned about cosets of a
subgroup in a group. We have learned about how cosets partition
a group and using that idea we’ve proved the first important
theorem of the course which was Lagrange’s theorem. And
remember that Lagrange’s theorem says that when that G is a
finite group and H is a subgroup of G, then the order of H
divides the order of the group.
(Refer Slide Time: 00:46)
So recall I am going to start with this, Lagrange’s theorem says
the following. So if G is a finite group, and H is a subgroup of
G, then we have the order of H divides the order of G. And in
fact we proved this by using counting formula, which is a more
precise statement about the orders of G or H. So not only does
the order of H divide the order of G we have this counting
formula, remember the symbol [G: H] stands for index of H in G
which is the member of left cosets of H in G.
So the ratio of the order of the group divided by the order of H is
precisley the number of left cosets, so this is a very important
formula and Lagrange’s theorem is a very important theorem.
And in order to explain how we use this, let me do two problems
in this video, so I want to illustrate the use of these results.

147
Problem one is let P be a prime number, and let G be a group of
order P. So P is a prime number and G is a group of order P.
Show that G is cyclic. So recall that a cyclic group is a group
which is generated by ,cyclic means generated by, a single
element. And I will recall this definition again in the solution, so
if a group has order a prime number,
(Refer Slide Time: 03:18)
then it is a cyclic group. So let us see how to solve this, so
because G has order a prime number and 1 is not a prime
number, since1 is not a prime number, G contains an element a,
which is different from E. This is because if G does not contain
any element different from E, then order of G is 1. Because G is
just the set consisting of E, but then order of G is 1, at the same
time we know order of G is prime number, so it cannot happen.
Now consider the subgroup H generated by a, so remember H is
all elements of this form: E, A, A, A square, and so on. And
because it is a finite group it will stop somewhere, and the
inverses of A will also be included in this, so I am not going to
write A power – 1 and so on, because some A power positive
integer will be equal ton A power -1. Now Lagrange’s theorem
says what? Order of H divides order of G.
Now order of G is a prime number. And order of H divides it.
But because P is a prime number, only numbers that divide a
prime number are 1 and that prime number. So order of H is
either 1 or order of H is P. Clearly order of H cannot be 1, why
cannot this happen, because H contains A and E which are
distinct, so H cannot be order 1 group, so this must be the case,
but if H is order P, and G is also order P, this means H equals,
but note that H is cyclic by definition. It is a group generated by,

148
it is a subgroup generated by a single element, so it s a cyclic
group by construction. And H is equal to G, so G is cyclic, that’s
all.
So we have solved the problem. If you have a group of order,
prime order it is cyclic automatically. You might wonder what
about groups that are not of prime order, so does there exist a
group of order 4 that is not cyclic. See, 4 is not a prime number
so the problem here is nothing to say about a group of order 4.
It however says that order of group order two must be cyclic, a
group of order three must be cyclic, group of order five must be
cyclic, seven, eleven and so on. But four we don’t know, does
there a group of order 6 is not cyclic?
(Refer Slide Time: 07:11)
We know a group of order 6 that is not cyclic. We know that
because we can take S3 which remember in our new notation
has this form. This has order 6 but elements have all order 1, 2
or 3. So this is, I will leave this as an exercise for you, it is a
very easy exercise, it’s something you should do, we have done
enough to conclude that it is not cyclic. But I haven’t explicitly
talked about a group of order 4 that is not cyclic, that I will
come back to later in the course. So this is anyway a digression,
the problem was to show that any group of order equal to a
prime number is cyclic. And we have used it becomes
immediate, the solution is immediate if once you use the
Lagrange’s theorem, Lagrange’s theorem is very powerful. The
point of the problem is to illustrate that.
One more problem I want to do, because to illustrate more about
cosets, so let us take a group homomorphism, phi is a group
homomorphism. Show that order of G, ok so again assume G is

149
finite, so whenever we talk about order we have to assume it is
finite. Show that order of G is equal to product of the order of
kernel phi and image phi.
Remember kernel phi is the set of elements G which map to
identity element of G prime. Image phi is simply the image of
the function. All element of G prime that have pre-images in G.
So show this, so this is also a problem which illustrates
important features of group homomorphisms, as well as
application of the notion of cosets.
So I am going to give a name for this, so let N equal kernel phi,
so recall that we defined group of homomorphisms, we showed
that kernel of a group homomorphism, and image of a group
homomorphism are subgroups. Not only that, after I defined
normal subgroups. I showed that kernel is a normal subgroup.
So we know that N is a normal subgroup of G, and we also
know that image, let H prime, the image of phi. Instead of
writing kernel φ, image φ all the time, I am giving those names.
Image φ is a subgroup of, H prime is a subgroup of G prime and
in general it is not a normal subgroup, as you should think of an
example.
(Refer Slide Time: 11:13)
So we want to show that order of G is equal to product of order
of N and order of H prime . So what I will do is, we will show
the following, before that I want to recall for you, if A and B are
in the group G, φ of A is equal to φ of B, if an only if, B is An,
for some n in N, why is the, this was done by us but let us
quickly recall this, so I will write it separately, φ of A is equal to
φ of B. That means φ of AB inverse, let me write this as A
inverse B is E G prime. This is because φ of A inverse times B is

150
φ A inverse, times φ of B, φ of A inverse φ of A whole inverse,
so these are inverses of each other and multiplying them by you
get identity element of G prime. I have done many examples of
this, so I am not going to write down the details. But this A
inverse B goes to G prime by definition A inverse B belong to N.
That means B belongs to AN. This is exactly the statement, and
note that if this happens, if B is in AN, A inverse B in N. If A
inverse B is in N by definition φ inverse A inverse B is E G
prime. And if φ of A inverse B is EG prime then φ of A must be
equal to φ of B. So these are implications, go backward also.
So this is the if and only if statement, this is the statement that
AN equals BN. So if B is equal to AN, A must be equal to the
left coset. Recall that AN is the left co set, A times small n for all
small n, BN is B times small n for all n. Remember that if B is
equal to An, An is an element of A capital N, B is an element of
B capital N, because you can take small n to the identity
element, so B is certainly in BN. So AN and BN have a common
element mainly b, hence they must be equal cosets. Remember if
two left cosets are either identical or they have nothing in
common. So we will use this to solve the problem. Now lets us
recall the problem.
We are trying to show that cardinality of G or namely the order
of G is equal to order of N times of order of H prime. So this is
the product of two numbers. So why is this true? I want to say
that I will take the set of, consider, so I will write it here, the set
of left cosets of N in G. This is denoted by, so I am introducing
new notation for you and this anticipates our definition of
quotient groups later, G mod N, that is how you read this, G
mod N is the set of left cosets, for now it is simply a set.

151
So consider a map, consider the function from the left cosets to
H prime, H prime you remember the image of phi. What is the
left coset? What is the function I want to consider? I take
element AN left cosets are of the form AN and I will map it to φ
of A, so this is the function. I am defining the function AN going
to this, this is the function F. So F of A is just φ A.
(Refer Slide Time: 16:07)
I claim that, we will show that F is bijection. What does this
mean? So first we want to show f is onto. Why? So you give me
any element of H prime, in order to be onto, a function has to
have a pre-image for every element in H prime. So you give me
an element of H prime. What is an element of H prime?
Remember H prime is the image of phi. So let φ of A be in H
prime. H prime is the image, so it only contains element of the
form φ (A) where A is in G. Then what is F of AN? F of AN is
precisely φ(A). By definition F of A N is the function which the,
F send AN is φ A. So if φ A is in H prime F of AN is φ A, so
every element of H prime is in the image. So F is on to, now we
will show that F is 1-1, what is the meaning of 1-1? If two
elements under F map to the same element, those two elements
are equal.
So suppose that F of AN is equal to BN, so take two cosets
which have the same image under F, this means if φ A is equal to
B, because F AN is by definition φ A, f BN is by definition φ B.
And then I checked earlier that if φ A is equals to φ B, then AN
equals BN. By the above argument, we conclude, because φ A
equals φ B implies AN= BN. So we have concluded that F is 1-1
also, F is onto, F is 1-1 also, we checked. So F is a bijection
between two sets, what does it mean? F is bijection between G

152
mod N and H prime.
If you have a bijection between two sets, the two sets must have
the same number of elements, the number of elements in G mod
N is equal to the number of elements of H prime, which is what
we denote by order H. But what is number of elements in G mod
N. Number of elements in GN is being the set of left cosets,
number of elements in G mod N is precisely the number of left
cosets of N and G, which we do not buy this symbol [G: N]. [G:
N] is equal to the cardinality of H prime. But what is [G:N] by
counting formula, which I recalled at the beginning of this
video, counting formula says G mod N, is equal to order of G
mod order of N, which is equal to order of H prime, which
implies order of G is equal to order of N times order of H prime,
as desired. This is exactly what we wanted, the problem asked to
prove that order of G is equal to order for kernel φ times of
order of image φ, kernel φ was denoted by N, image φ was
denoted by H prime, so order of G is equal to order of kernel φ
times order of H prime, so that completes the solution.
So in these two problems I hope you understood how to use the
counting formula and Lagrange’s theorem and using these we
are able to conclude very strong statements about groups, so I
will stop the video here and in the next video I am going to start
with my definition of quotient groups.
Online Editing and Post Production
Karthick
Ravichandaran
Mohanarangan
Sribalaji
Komathi

153
 Vignesh
Mahesh kumar
Web Studio Team
Anitha
Bharathi
Catherine
Clifford
Deepthi
Dhivya
Divya
Gayathri
Gokulsekar
Halid
Hemavathy
Jagadeeshwaran
Jayanthi
Kamala
Lakshimipriya
Libin
Madhu
Maria Neeta
Mohana
Mohana Sundari
Muralikrishnan
Nivetha
Parkavi
Poonkuzhale
Poornika
Premkumar
Ragavi
Raja

154
 Renuka
Saravanan
Sathya
Shirley
Sorna
Subash
Suriyaprakash
Vinothini
Executive Producer
Kannan Krishnamurty
NPTEL Co-ordinates
Prof. Andrew Thangaraj
Prof. Prathap Haridoss
IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resource Development
Government of India
www.mptel.ac.in
Copyrights Reserved

155
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 04
Lecture 19-“Quotient groups”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE
(Refer Slide Time 04:20)

Okay, so in the pervious video we have seen how to apply Lagrange’s theorem and
counting formula, and before that we have seen how to define cosets of a subgroup in
a group. And in this video I want to introduce to you a very important notion of group
theory namely quotient groups, okay this is a very important notion that is used all
the time in group theory. So it is important for you to carefully understand the notion
and be able to apply various theorems and properties of quotient group that we will
study. So I am going to introduce to you what quotient groups are using some
definitions and some properties.
The basic idea is the following you have a group, G and H a subgroup of G, in the
previous video I introduced for you the notation G mod H, it was this side of all left
cosets of H in G so in other words these are all elements of the form AH where A is in
G, to begin with this is just a set and we computed this set in some case for example
when G is the group of integers under addition and H was some subgroup of the form
NZ and I think we worked out the case of 4Z in our example the set of left cosets was
the finite set actually and we listed them. Similarly when G was the group of the
bijections of the three element set namely symmetric group, on three letters S3 and
we looked at some subgroup H and we wrote down all the cosets of that subgroup. So
this is the set of our left cosets so our question is that will motivate the definition of
quotient groups is: can we give a group structure to G mod H? So this is the basic
question I want to address. Can we give a group structure to the set of left cosets of H
in G, so in other words what is a group structure? It is a first and for most it is a
binary operation so, given AH and BH in G mod H can we do perform something,
can we get after performing some operation so, third left coset out of AH and BH? So
this is the question.

156
So question is can you give a group structure to G mod H and in order to do that first
we want to define a binary operation on G mod H so that when you apply that two
cosets AH and BH we get a third coset. We want to do this in general now a specific
example you may able to do it, but the hope is that we should be able to do this in
general and not only of course this is not the end of it right we have to after giving
this binary operation we have to make sure that there is a identity element that this
group operation admits inverses and it is associative, these must be checked.
(Refer Slide Time 04:26)

So let us see whether we can do something like this so, I am going to recall for you
the example we worked out. So lets take G to be the symmetric group on three letters
and H to be the subgroup of E and (1,2) and if you recall we computed the cosets of
this very carefully. So in particular we computed what is (1,3)H. (1,3) H was so I
want to just write this it is something that we already did. And let me work with
another element (2,3) H, (2,3) H was (2,3) and (1,3,2) we already did saw this in a
previous video we have done this. So what do we do can we combine them, so what
is (1,3) H what should be (1,3)H times (2,3)H? So it looks like we have to learn how
to multiply sets, we know how to multiply elements, so one attempt would be, so now
very generally, let me introduce to you a notion: G is a group. A,B or subsets of G not
subgroups okay, so I am not taking not subgroups not necessarily subgroups. They
can also be subgroups but I am not restricting only to the case of subgroups. Define
because I am doing this because it seems like we have to multiply two subsets of a
group here in order to multiply 1,3 H and 2,3 H we have to multiply two subsets. So
let us define A dot B to be simply a times b where a is in A and b is in B, this is just
the product set. Very natural definition right, in order to multiply two sets we simply
take elements of the first set multiply them with the elements of the second set and
you vary all the elements. So can we use this definition because cosets remember are
just subsets, left cosets are nothing but subsets and I told you how to multiply subsets
of a group. So using this to calculate the two cosets we are considering here (1,3) H
times (2,3) H, I need to remind you again I am not claiming this a group operation
this will work out, I am just saying this is an attempt. Because we need to multiply
here we have to multiply sets, the goal is to multiply sets. So if you want to multiply
sets, this seems like a natural definition of set multiplication,
(Refer Slide Time 08:04)

So lets just do (1,3)H times (2,3) H and as I said (1,3)H and (2,3)H are we listed them
earlier, it is simply the set (1,3) lets see 1,3H was (1,2,3) product with (2,3) (1,3,2) so
these are the two left cosets. So lets multiply them, you take 1,3 and multiply by you
taken an element from the first set in my definition I take this to be A and take this to

157
be B, this becomes 1,3 times 2,3, so (1,3) times (2,3), (1,3) times (1,3,2), (1,2,3)
times (2,3), and (1,2,3) times (1,3,2) okay. Those are the four products we have so I
take the fist element of A 1,3 multiply by 2,3 because that is an element of B, I take
1,3 and multiply it by 1,3,2, similarly I take 1,2,3 multiply by 2,3, 1,2,3 times 1,3,2,
so these are the four products. Let us just simplify these and see what we get, what is
(1,3) times (2,3) so I will just quickly do one these and write the rest. Remember the
product in the new notation for S3 how do we multiply, so we will start with the right
hand side element and go to the left. So 2 goes to 3, and 3, goes to 1, so this 2,1 what
does 1 do, 1 is fixed under 2,3 and 1 goes to 3, so this is (2,1,3) 3 goes to 2, and 2
goes to 2 in this side so this is (2,1,3) similarly 1,3 times 1,3,2 is, 1 goes to 3, and 3
goes to 1, so 1 is fixed, 2 goes to 1 and 1 goes to 3, so 2 goes to 3, and 3 must go to 2,
as you can check 3 goes to 2, and 2 goes to 1, similarly (1,2,3) times (1,2,3), 1 goes to
2 because 1 is fixed here and 1 goes to 2, 2 goes to 3 and 3 goes to 1 so 2 goes to 1, 3
goes to 2 and 2 goes to 3 that is fixed. Finally 1 goes to 3 and 3 goes to 1 so 1 is fixed
2 goes to 1 and 1goes to 2 so 2 is fixed, and 3 goes to 2 and 2 goes to 3 so this
actually the identity element. So the last product is the identity element and this of
course can be re-written as (1,3,2 ) so this another subset, so I am just going to
rearrange these elements. E, (1,2),(2,3),(1,3,2).
So now lets stare at this earlier I taken, I have defined taken two subsets and define
the product it is another subset, so here I have taken two subsets, consisting of two
elements each and I multiplied them I got a subset of four elements. Now in order for
this operation to be a valid group operation on the set of left cosets, when I multiply
left cosets I must get another left coset, so is this a left coset? We computed the left
cosets of this subgroup in the earlier video and either by looking at the list of cosets
that we computed or using counting formula. Counting formula say the number of
left cosets, elements in a left coset are equal to the elements of H, remember
cardinality of H is equal to cardinality of AH, for every A in G, so any left coset of H
must have same number of elements as H, but AH as two elements in our example so
any left coset must have two elements, but this has four element so this cannot be a
left coset. You can either check the explicit list that we calculated earlier or use the
counting formula, not the counting formula really it is the theorem that cardinality of
H is the same as cardinality of AH, this was used in Lagrange’s theorem and counting
formula.
So this cannot be a left coset, so this too bad, we cannot use this definition to give the
definition of a multiplication of two left cosets. And however I want to now illustrate
that it is actually not such a bad definition of product of left cosets, we only need to
ask for a slightly stronger condition on a subgroup. We need to impose condition on
subgroups,
(Refer Slide Time 13:14)

158
so the conclusion is this product of left coset is not a valid binary operation because
product of two left cosets is not a left coset in general. So in general product of two
left cosets is not a left coset, so it is not a valid binary operation.
(Refer Slide Time: 14:07)

However we will show in the next few minutes, this is a valid binary operation if H is
not just a subgroup, H is a normal subgroup. If H is a normal subgroup, it turns out
that this product we define here A * B = { a b / a € A , b € B } that becomes a valid
binary operation provided that H is a normal subgroup, ok, this is what I want to
show. But before that let me prove a proposition to achieve my goal to show the
following.
Suppose that H is normal in G so, typically we denote normal subgroups by N, so in
this case I will stick to H, but subsequently, when we talk about normal subgroups we
will use the letter N. So what I want to first show is if A belongs to G then A H = H
A, so if H is normal in G, AH=HA, in other words the left coset AH is equal to the
right coset HA. Let us prove this, this is not difficult.
So I am saying in fact I am saying that if H is normal AH=HA, I am also saying the
converse but I will write that later. So I am saying AH=HA, what does that mean? I
want to prove that, the set AH is equal to the set HA, what does this mean what is
AH? We want to show that AH as H is in H equal to HA as H is in H, so an important
warning for you. We are not saying that Ah=hA, in other words if I replace capital
(H) by small (h) here, I cannot say this, see this is too much to expect. This is saying
that, if you recall the definition of, if this is happen for every A, this mean that H is in
the center of the group if you recall the definition of center of a group, but H can be
normal and not be in the center of a group. So I am not saying this, why am I not
saying this, this is a weaker statement, because I am saying that AH A small h is equal
to Ah small h times some other small h times A, so we only need AH to be in HA, it
need not be this particular small h times A, it can be some some other h times A, so it
is very important to keep in mind that to say that these to cosets are same is not same
as saying A, the element A time small h is equal to the element small h times A, so
now lets prove these two sets are equal, how do in general prove two sets are equal?
You want to show that the first is the subset of the second, and the second is the
subset of the first.
(Refer Slide Time: 18:17)
So to prove this equality choose or let Ah be in AH, so if Ah in, A capital H, lets
rewrite like this write Ah is equal to Ah(A inverse) A, we can write it like this right,
because by associativity A inverse and A can be first multiplied that is identity and
you have Ah, so but then since, H is normal, H is normal in G, this is where we are

159
using the normal hypothesis given A in G, and h an H, A h A inverse belongs to H,
definition of normal subgroup, so this is H, I am calling it as H prime, H prime is in
H, this is my definition in H here, so if you take an arbitrary element of the left coset,
it is in the right coset. So AH is a subset of HA, and opposite is identical, you can do
this by symmetry but let me just illustrate the point again if it is not clear to you
hopefully it will be clearer after this, lets take an arbitrary element of the right coset.
Then what we write, as before let us write it as A times A inverse times hA, that is
same as this because of the associativity. But if Ah A inverse is in H, this is also in H,
so this AH, so called this equal to H double prime, AH double time, and this is by
definition of a left coset is an AH, so every element here arbitrary HA, is in AH, so
HA is in AH, hence AH equals HA, so this completes the proof of the proposition.
Proposition only asked you to show if H is normal AH equal HA for every A.
(Refer Slide Time: 21:03)

I want to leave this an exercise for you and this is a easy exercise that you must do: if
H is the subgroup of G, and AH equals HA for all elements of the group then H must
be normal in G, so if this property holds for every element then it is actually normal.
That we do not need, so I will leave it as an exercise for you but you should do this
and actually another exercise. If G is S3 and H is the subgroup we considered earlier E
and 1,2, is (23) times H equal to H times (23)? Okay so this is the question left coset
of 23 is equal to right coset of 23? So I will give these exercises to you, you should
do this, this 23 H we already computed just compute H 23 okay, this if you compute
you will see that then you may ask if it is equal or not.
(Refer Slide Time: 22:19)

So now lets come back to normal subgroups, so in a normal subgroup we have AH

equals HA now I want to prove the main result that we want. Proposition: Suppose
that H is a normal subgroup of G, then aH, then let A,B be in G, so I am taking two
arbitrary elements of G, AH times BH is by definition, this is the product of two sets
that I have defined earlier. So this is the product of these two cosets that we defined
earlier, namely you take something in AH times something BH and multiply, you do
this for every element of AH and every element of BH, and we saw in the example of
S3 and H equal to E and 1,2 that the product is not a coset but in this case it becomes a
coset. It is a coset actually AB times H, so if AH and BH are two cosets their product
is also a coset. Lets prove this, this is not difficult now.
(Refer Slide Time: 23:20)

160
So we already know by the previous proposition, what do we know, so we know that
HB equals BH, so let me write that here HB equals BH, so now lets compute this, AH
times BH, see, remember just to illustrate, so will just start in the new page.
(Refer Slide Time: 24:33)

AH times BH, because of associativity, I can write this as A times HB times H, so

why is this? It might be useful, it is true but it might be useful to write down this,
what is AH times BH, you take something in the left coset AH it is of this form, and
you take something here, so BH prime, so you take this as you vary H, in H and H
prime in H, this exactly the product of two set in a group that we defined earlier. If
you have two groups, subsets of a group, the product is defined to be A,B were A is in
A, and B is in B, this is the definition of product of subsets that we will always use.
So in this case I applying this definition to AH and BH, AH is the set A small h, BH is
the set B small h prime, I just want to use a different letter because it is you have to
vary all elements of capital H, so AH times BH, but this, if you associatitviy of group
product, then this becomes, so I can vary a H and H prime two distinct element to
may not be distinct two elements of H, then I have AH, B H prime but I put the
bracket around HB, but this is precisely this, because HB, capital HB is all element
small hb, and capital H is small h prime, so these two are equal, so this is okay, these
two are equal.
But now by the same associativity I can write this as, first of all use the property I
said earlier. HB equals BH so I can interchange, remember I cannot interchange B
and small h, but HB equals some BH double prime, so that BH double prime will
come here. So in other words I am saying that A,B,H double prime, H prime, H
double prime H prime in H, so this part can be interchanged, switched like this. So
this is what I have now again using the same this is this, again using the same
associativity I can write this as AB times HH, but what is HH, if you think about it, it
is nothing but H. So we claim HH equals H. Why? What is HH, it is the product H
times H prime, H and H prime are in H, right, so but if H and H prime are both in H,
HH prime is in H, so this is actually contained in H, HH is contained in H, on the
other hand if small h is in capital H, you can write this as H times E and this is in HH,
because E is in H, identity element is in H, H is in H so HE is in HH, so this means H
is contained in HH, so HH is contained in H, H is contained in HH, so this implies H
equals HH, when you multiply a group with itself you do not get anything bigger.
You get exactly the group this is not true in general for is subset only which is not a
subgroup.
So H is equals HH, because H is a group, we have used both properties here because
if H and H prime are in H the product needs to be in H here, and E is in H, we need,
E in order for H, H is equal to HE right, but for that we need E to be in H, so because

161
we have used the other important property of a subgroup, it must contain the identity
element. So, see now we are done because AH, BH is AH, BH, which is BH, H which
is AB, HH, which is AB, H, so AH, BH is equal to ABH, which is what we claimed.
So this is the theorem, so this proposition allows us to define a group structure on G
mod H, when H is normal. It turns out that there is no natural way to define group
structure on G mod H, when H is not normal. This can only be done when H is
normal, so what is the group structure, so from now on assume that G is a group and
H is a normal subgroup. So I am going to assume that always, I am dealing with a
normal subgroup of a group G, then we will give a group structure to G mod H.
So first of all, recall what is G mod H? This is the set of left cosets, as you vary A in
G, so what is the binary operation? Remember the definition of a group is you start
with a set and define a binary operation on the set and verify that there is an identity,
every element is as an inverse and the binary operation is associative. So what is the
binary operation? It is the one we discussed earlier. So define AH*BH, I am using
star just to emphasize that it is a new binary operation we are defining, is the product
ABH. So this is the definition of the binary operation. Now is there an identity
element? It is simply H, see H itself is a left coset right, that is identity. But why is
that? You take AH*H, and this is AH because A*E, is A, similarly H*AH is AH. so
this is an identity element. What is the inverse of AH inverse clearly must be A
inverse H, because AH time A inverse H is, by definition A, A inverse times H, which
is EH which is H. Similarly A inverse times H times AH, is A inverse AH, which is
EH, which is H.
(Refer Slide Time: 32:50)
And associativity finally. If you take AH times BH, then take CH, this is AB H times
CH which is by definition AB, CH remember A,B,C you can multiply in any order
because this is happening in capital G, so in group G the product is associative. See,
as you can see, I am not writing star so if there is no star I just multiplying using this
definition, so AB, CH. On the other hand what is AH, times BH times CH, this is AH
times BCH which is really what I have to write here is AB times C, here I have A BC
times H, but these are equal because AB times C, A times BC, so associativity is also
clear.
The point is you needed a valid binary operation, that was the key step, if you have a
valid binary operation it turns out the other are more or less is clear, identity element
is EH inverse is are given by inverse of these elements, associativity is essentially the
associativity in the group itself that will give associativity of G mod H.
You might wonder, can I can just define this without H normal. As I said if you just
declare it like this, many of the properties you cannot check because AH*H would be
the product, is no longer going to be AH, if H is not normal. So if you just declare it
without the normality hypothesis this will not give you a group. So this is a quotient

162
group that we wanted to discuss in this video ,G mod H with this group structure, is
called the quotient group.
So the quotient group is defined, this is defined, only when H is a normal subgroup,
so and quotient groups come with an important homomorphism. There is a natural
homomorphism from G to G mod H. What is that? You take A and you simply map it
AH, this is the set map, A going to AH because G mod H remember is the set of left
cosets, A is an element, AH is the left coset. Why is this a group homomorphism?
Why is phi a group homomorphism? Let’s check this, we want to check phi of AB
equals phi of A times phi (B). Is this true, lets write down, what is phi of (AB), phi
(AB) is by definition (ABH) right and phi of (A) is (AH) phi (B) is (BH) and these
two are equal by definition of the product. (AH) times (BH) is equal (ABH), so this a
group homomorphism. So quotient groups must always be thought of with this
structure of this particular group homomorphism, so it is a group and it admits a
natural homomorphism always G to G mod H. What are the properties of G mod, phi,
we can say the following. Kernel of phi, remember kernel of phi is all A in G such
that (AH) is equal to the identity element of G/H, right, this is by definition the
kernel, what is the identity element of G/H, so these all are A in G such that (AH)
equals (EH) which is same as (AH), so I don’t need to write EH, I will just write H,
the identity element of G mod H remember is H, so it’s all elements that map to H,
but what is this? When is AH equal to H this happens precisely when A in H, this is
something we have discussed earlier, so this is H, kernel phi is H. What is image phi?
It is AH image phi by definition phi (A), A in G. That means this AH with A and G,
this is precisely GH, G mod H. You have all the left cosets here, so its G mod H, so
in other words phi is onto, so kernel phi is H and phi is onto, so I am going to
summarise all this, that we learnt about quotient groups and write the following.
(Refer slide Time: 39:09)
If G is a group and H is a normal subgroup of G, then the set G mod H of left cosets
of H in G is a group under the operation AH times BH equals ABH. This group G
mod H is called the quotient group, it is the quotient group associated to G and H. It
comes with a natural group homomorphism phi from G to G mod H such that, so we
have checked that phi is a group of homomorphism, such that phi is onto and kernel
phi is H. So this entire thing you need to remember, so read this carefully we can
construct quotient groups whenever you have a group and a normal subgroup and the
underlying set for the quotient group is simply the left cosets of H in G, so I want you
to think of quotients groups as something very easy, so it is not difficult at all,
because you know what is the set of G mod H, it is the simply a left cosets. A group is
a set with binary operation right, so you know the set here it’s just G mod H, it is the
set of left cosets, what is the operation? We need to give it an operation to make it a
group, it is simply you take AH and take BH multiply them, you get ABH, so that is
the group operation, and it’s easy to check that this group operation is closed, has
identity, has inverses and it is associative. Not only that, quotient group must always

163
be thought of along with this crucial natural group homomorphism, from G to G mod
H, which has two important properties, namely that it is onto and its kernel is H, so
these are the properties of quotient groups that we want to emphasise.
Online editing and Post Production
Karthik
Ravichandran

164
 NPTEL

NPTEL ONLINE COURSE

Introduction to Abstract

Group Theory

Module 04

Lecture 20-“Examples of quotient groups”

PROF. KRISHANA HANUMANTHU

CHENNAI MATHAMATICAL INSTITUTE

(Refer Slide Time: 01:51)

Now let me go ahead and do some examples, to better understand

this notion. I am going to take the standard example, the main
example that we want, we have learned in beginning namely the
group of integers, so this is under addition, actually this is an
abelian group right, these G is abelian so, every subgroup is
normal, remember normality is a trivial property once you have an
abelian subgroup, abelian group, because AB A inverse, what is
normal mean? Means if A is in H sorry A is in G, H is an H, you
want AH, A inverse is in H, but if you have an abelian group, this
is trivial because A, H G is abelian implies AH, A inverse is a A A
inverse H, which is H, that is certainly in capital H, so if you have
an abelian group, every group is normal, that is trivial, so we can
take any subgroup. In the previous slide we had,

(Refer Slide Time: 01:53)

165
H needed to be a normal subgroup, because now I am working
with abelian group, and everything is normal, so I take any
subgroup I want. For example, let’s take 2Z, remember subgroups
of integers are nZ, where N is an integer, so let me start with 2Z.
So what is G mod 2Z, G mod H, this is same as Z mod 2Z, so
what is this? So again as I said, you have to believe that quotient
groups are easy to understand okay, they are easy to understand,
you systematically work out what needs to be done.

When I ask you what is Z mod 2Z, you should ask to yourself,
what is the set first, what is the underlying set? What is the
underlying set? Before you answer what group it is, you want to
know what set is it, once you understand the set you can then look
at the binary operations. What is the underlying set? It is simply
all left cosets of 2Z and in Z, right, what are they? So certainly we
have 2Z, remember left cosets are, in this case I am working with
the group of integers, so in this case, the operation is addition, so I
am going to denote cosets buy M plus 2Z, so these are all the left
cosets. As you vary m, are the all the left cosets, right, now though
it looks like we get one for each integer as we are seen in the
previous videos many of them collapse, many of them are
identical, for example you take zero plus 2Z that is same as two
plus 2Z, right, that is same as four plus 2Z, that is same as six plus
2Z, that is same as minus four plus 2Z, that is same as 102 plus
2Z, and so on. Okay in fact, if m is even, this is equal two m plus
2Z, for any even integer m okay.

(Refer Slide Time: 04:38)

If, this is of course 2Z, m plus 2Z is equal to 2Z, for every even
integer, similarly one plus 2Z is equal to m plus 2Z, for every odd
integer okay. This is because, we have done many times, so I do
not want to spend a lot time but 2Z remember, is all even integers,
if you add one to this, you get minus three, minus one, one, three,
five, and so on, if you add two, three to this okay, you get minus

166
one, minus four plus three is minus one, minus two plus minus
three, sorry minus four minus one, minus two plus three is one,
then you get zero plus three is three, two plus three is five and so
on.

So it as a set, these two are equal right, because this is just shifting
that zero, and zero plus one is one, minus two plus three is three,
so these two are equal, so in fact I mean this is a quick explanation
of why, one plus two Z is equal to m plus 2Z, every odd integer
would be contained in this set, so this is in fact, so in other words,
2Z is the set of even integer, one plus 2Z is the set of odd integers.
What you have to focus on is, when you divide by two, what is the
reminder? It can either be zero or one, if it is zero it is an even
integer and if it is one, and it is an odd integer,

(Refer Slide Time: 06:46)

So again let’s come back to our main focus here, I am trying to

understand what is the Z mod 2Z, as a set it is simply 2Z and one
plus 2Z, okay as I said again remember that quotient groups are
very easy to understand, because first now understood the set and
next is to understand the operation, what sometimes confuses is
people is the way that we describe the elements here, one is 2Z,
and other is one plus 2Z,

But do not worry about it, it is a actually irrelevant what we call

this, so let me call them E comma A okay, so E is 2Z and A is one
plus 2Z, I am going to rewrite this in our standard notation of
groups, 2Z was the identity element and I am going to call it as E,
the other element is just called A, so when you write it like this, it
becomes fairly clear, right, it is just a two element set which are all
even integers and all odd integers, don’t think like this if that
confuses you, think them as E comma A, okay this is the set and
what is the operation?

167
(Refer Slide Times: 08:10)

There are only two elements here so, it is very easy to define the
binary operation, so, what is 2Z plus one plus 2Z, so again I am
going to rewrite this as in my simplified notation it is just E times
A, so I am going to switch the additive notation to product
notation, what is this? This is if you recall the definition of the
quotient group here, AH and BH simply was ABH, where AB is
the group operation of the group G, in this case this is Z, it is zero
plus Z, zero plus one plus 2Z, this is 2Z and this is not surprising
because A time E must be A, similarly what is one plus 2Z plus
one plus 2Z, all I do is first I add one and one and I have 2Z, that
is same as 2Z, in the notation that I want you to think one plus 2Z
was A, right, this is A and this is E, this A squared equals E, okay
so you have a group that’s it you have described the group
completely,

(Refer Slide time: 09:39)

So Z mod 2Z is a group with two elements E, A and the

multiplication table remember multiplication tables are given by,
you have E,A and E,A and what is EE? That is E, this is A it is A
and this is E, that’s all, so do you agree Z mod 2Z is a very simple
group, it is just a group with two elements with the multiplication
table given by this. Note that quotient group are really not just
groups but you need to always think of the group homomorphism
from the group to the quotient group, and what is the natural group
homomorphism? Z to Z mod 2Z, so remember that we have a
natural homomorphism always, the part of,

(Refer Slide time: 10:58)

this are summary of this very important, not only do we talk about
G mod H being a group, we come always, we also discuss the
natural group homomorphism from G to G mod H, which is onto

168
and has the kernel equal to the subgroup H, so whatever it is in
this case

(Refer Slide time: 11:17)

it must have kernel equal to 2Z, but it is clear now right, what is m
phi m is by definition m plus 2Z, so this is, in my notation this is
E, if m is even, because remember I am identifying Z mod 2Z as
E,A so this is E when m is even, and A if m is odd. So this is the
group homomorphism, it is certainly onto because all even
integers map to E and all odd integers map to A, so everything in
Z mod 2Z is in the image.

And what is kernel of phi, things that map to identity m in Z such

that phi m is E, so this is even integers, this is precisely 2Z, okay
so we have in this very simple example Z mod 2Z, we have
worked out the left cosets, we have worked out the group structure
and you have to, remember because it is an abstract group theory
course, you have to abstract out the crucial piece of information, I
do not care what the form of elements are, 2Z, one plus 2Z that is
not very important, so to simplify I am just going to call them E
and A, and the multiplication table completely describes the group
structure.

(Refer Slide time: 12:58)

And quotient group is given by this group homomorphism, which

also we completely understand, so if you understand each piece of
this game here, you know what the quotient group is completely
okay.

(Refer Slide time: 13:12)

169
Now I will just, for more exercises for you, to make sure that feel
comfortable with group quotients, let’s take 6Z, so what would be
or rather let me say describe Z mod 6Z. Now I am going to do this
quickly because we have done it in full detail for Z mod 2Z, what
is the set? First question as I said what is the set? So what are the
lest cosets, okay I am not going to describe this in all detail
because we have done it for Z mod 2Z, so quickly we will
understand, I hope you will agree with me he quotient groups are
easy, okay, it is very easy to describe quotient groups, let’s first
start with, what is the Z mod 6Z, so in other words, what are the
left cosets, let me remind you that 6Z is a normal subgroup of Z,
so we can talk about the quotient group.

And Z is an abelian group, so any subgroup can be taken, what are

the left cosets? There are actually six left cosets, just like there
were two left cosets when you were working with the 2Z, there
will be 6 left cosets, why is this? Briefly the reason is, you take an
integer, you divide by 6, it will have a reminder which is between
0 and 5 and that will determine where it is, which coset will
contain it. So this is 6Z, one plus 6Z, two plus 6Z, 3 plus 6Z, and
four plus 6Z and five plus 6Z, for example you take 25 okay 25 is
an integer. When you divide by six, this is 6 times four plus one,
so 25 will belong to this, so 25 is an element of this, what about
56, 56 is when you divide by, this is nine times six plus two right,
nine time six it is 54, so this is inside 56, so every integer is in one
of them, they cover, they partition the group, the group Z, and any
other coset is equal to one of them, for example this means 56 plus
6Z is equal to two plus 6Z, so the quotient group Z mod 6Z has 6
elements and again it might be confusing to keep track of these
particular descriptions, I do not need to do that, I am just going to
simply take E, A1, A2, A3, A4, A5, E was 6Z, AI is I plus 6Z.

(Refer Slide time: 16:42)

170
So this one plus 6Z and two plus 6Z and so on, what is the group
structure? So what is A1 times A2, I am not going here the
multiplication table is going to be six by six, so let me not write
the full thing but for example what is the A1 times A2, A1 times
A2 is, one plus 6Z plus 2 plus 6Z, that is three plus 6Z, so that is
A3, right, so what is A1? So I worked out A1 times A2 let me
work out what is A1 times the A1, that is one plus 6Z, plus one
plus 6Z, so that is two plus 6Z, that is A2. What is the A1 times,
A1 cubed, this I A1 squared and what is A1 cubed?

This is A1 times A1 times A1, so this is one plus 6Z plus one plus
6Z plus one plus 6Z, and this if you think about It is three plus 6Z
and that is A3, what is the A1 power four? So this is A1, A1,
A1,A1, so this one plus 6Z, one plus 6Z added four times this is
4+6Z that is A4. What is A1 power five? That is A1, A1, A1 A1,
A1, one plus 6Z plus 1 plus 6Z six times sorry this is five times
you know that I should write A5. And finally what is A1 power 6?
A1 six times so this is A6, but what is A6? There is no A6. One
plus 6Z plus one plus 6Z, six times. That means you get 6+6Z.
What is 6+6Z? Remember we have only 0+6Z up to 5 +6Z. 6+6Z
if you think about it is just 6Z. So this E. So A1 power 6 is E.

(Refer Slide Time: 18:59)

So what is the subgroup generated by A1? So this E, A1, A1 2, A22,

A32 sorry this is not correct it is A1 2, A1 3, A1 4, A1 5 and then
A1 6 versus E. So this is what it is. And this is actually because of
the calculation we have made it is E, A1, A2, A3, A4, A5 so this is
just Z mod 6Z. Any group which has an element like this has a
name for it. What is that name? So, Z mod 6Z is a cyclic group.

This means at this point we can say order of A is 6. So 6 is the

smallest positive integer such that A1 power 6 is identity. Okay Z

171
mod 6 is a cyclic group of order 6. Okay so this is just a
observation we made but the point is we have understood Z mod
6Z completely. Let me also write down the natural homomorphism
from Z to Z mod 6Z. So where does phi of m go? It will go to m
plus 6Z.

Okay and it is equal to one of these. It is equal to you write m as

6a, 6 times p plus r. So r is the remainder. So this will be and r
remember will be between 0 and 5, when you divide by 6, the
remainder is between 0 and less than or it is greater than, it could
be 0, it could be 5. So it is between 0 and 5.

So it is r plus 6Z. So this is the, again remember phi is onto and

kernel of phi is 6Z. All the things where you, in order for you to be
in the kernel, r must be 0, meaning when you divide by 6 the
remainder must be 0 that means it is a multiple of 6. Okay, so what
we have concluded is Z mod 6Z is a cyclic group and it has a
natural map from Z.

(Refer Slide Time: 22:01)

As an exercise I will leave this for you. After we worked out these
two examples, for any integer n, Z mod nZ is always a cyclic
group, for any integer n, of order n, exactly n elements. So it is the
cyclic group of order n, okay so again you have to remember how
to list the cosets? So I will let you do this, I will not do the
exercise and it is a good exercise to do, to make sure that you
understand the notion of a quotient group.

I will only give you a hint which is to consider the left cosets of
nZ in Z which again depend on what is the remainder when you
divide by n. So you have possible remainders are 0, 1, and 2 up to
n-1. So these are the cosets and the product is determined by the

172
standard formula a+nZ+ b+nZ is, these are two left cosets, so it is
a+b +n Z.

Okay so this is the product or the addition, binary operation and

you can check that this is a group operation and it is a cyclic group
because 1+nZ will generate just like here, remember A1 generates
it that means A1 is 1+nZ, A1 was 1+nZ, A1 was the generator. So
we use that to show that Z mod nZ is a cyclic group.

(Refer Slide Time: 23:54)

So, let me give you one important point here. So recall that we
have defined the quotient group as follows. So G is a group H is a
normal subgroup. In this setup we look at the set of left cosets and
I defined (aH) remember I defined it like this. We have to address
the problem of whether this is a well-defined group operation,
because even though we have checked all the group properties
hold we have not really addressed this problem. Is this is the well-
defined group operation?

We need to, why is this important question? Because remember a

coset (a H) can be equal to another coset (a prime H) and A and A
prime could be very well distinct elements. Similarly, suppose that
bH equals b prime H. So, aH equals a prime H, bH equals b prime
H. Then in our definition we said aH times bH whatever it is
should be equal to a prime H times b prime H. Right, this must
happen in order for a valid group operation, a well-defined group
operation because this is a single element in that set that is equal
to this. Similarly, this is another single element.

That is bH and b prime H are actually same elements of G mod H.

So, when you multiply aH and bH it must be the same as a prime
H and b prime H and this is very important. I will give you an

173
example let us take G to be S3, that we have discussed in detail
and H to be E and 12. Right, we have seen this example in detail.
So, in this example let us try to compute, let us say I define, so
suppose we just define A H times BH equals AB H. What happens
if you just define it like this?

(Refer Slide Time: 26:42)

Let us look at 23H and we know 23H, if you remember the

calculation we have done 23H is same as 132H, this we know. So,
if you square this element, meaning you multiply it with itself,
23H let us use this description, 23H times 23H right, then in this
description what do you get? You get 23 times 23H but 23 is an
element of order 2, so 23H is same as eH, H right, but on the other
hand we can also use the description 132H. So, 132H is same as
132H, but this is, if our definition we apply, is same as 132 times
132H. What is 132, 132? If you do the product it is simply 123H.

But this is the problem. H is not equal to 123H. Right, H is not

equal to 123H. This also we see 123H is a different coset. So, this
is not a valid group operation. If you simply declare aH times bH
is equal to abH this is not going to give you a valid group
operation because 23H is same as 132H but you do it in two
different ways, you get two different cosets. So, it is important to
note that we need H to be normal. Then why the problem does not
happen? Why the problem does not occur?

Because here there is no problem, because we have shown in the

above proposition that the product of aH and bH, see this is not
the binary operation, this is just a set theoretic product of two
subsets of a group, is equal to right, the product of aH and bH is
equal to abH. Now if aH is equal to a prime H and bH is equal to b
prime H these are the same subsets right, these are the same
subsets so this product is also going to equal, at the same time it
will equal a prime b prime H, this the proposition but because

174
these two sets are equal these are equal. So, if H is normal then
there is no problem because aH and bH as a set, two sets their
product is abH. Okay, this is something that we have proved. This
is the crucial proposition that we have proved.

So if we have another description, it is just another name for the

set but the set is same aH is equal to a prime H because that is
only, I mean we are trying to address the problem if aH is equal to
aH, a prime H. In this case there is no problem, in earlier case
there is a problem because these cosets are same but in our
definition we are arbitrarily defining aH and bH to be aH times bH
to be abH. Why should it equal a prime b prime H? And sure
enough, as our example shows, they are not equal.

However in the proposition we showed that if H is normal the

product of the set aH with the product, with the set bH is abH and
if the set aH is equal to a prime H and the set bH is equal to b
prime H, we have no problem because if you have two sets the
product is what it is. Just changing the name of the set does not
change anything.

The product is the same. So, in the case that H is normal there is
no problem. The group operation is well-defined and we have
already checked that all the other properties hold and it becomes a
group. Okay, so do these exercises and this will help you
understand the notion of a quotient group and in the next video we
will look at more examples and look at more properties of quotient
groups. Thank you.

Online Editing and Post Production

Karthik

Ravichandran

175
 Mohanarangan

Sribalaji

Komathi

Vignesh

Mahesh Kumar

Web Studio Team

Anitha

Bharathi

Clifford

Deepthi

Dhivya

Divya

Gayathri

Gokulsekar

Halid

Heamvathy

Jagadeeshwaran

Jayanthi

176
 Kamala

Lakshmipriya

Libin

Madhu

Maria Neeta

Mohana

Mohana Sundari

Muralikrishnan

Nivetha

Parkavi

Poonkuzhale

Poornika

Premkumar

Ragavi

Raja

Renuka

Saravanan

Sathya

177
 Shirley

Subash

Suriyaprakash

Vinothini

Executive producer

Kannan Krishnamurty

NPTEL Co-ordinators

Prof.Andrew Thangaraj

Prof. Parthap Haridoss

IIT Madras Production

Funded by

Department of Higher Education

Ministry of Human Resource Development

Government of India

HYPERLINK "http://www.nptel.ac.in" www.nptel.ac.in

Copyright Reserved

178
 Introduction to Abstract
Group Theory
Module 04
Lecture 21 – “First isomorphism theorem”
PROF. KRISHNA HANUMATHU

CHENNAI MATHEMATICAL INSTITUTE

So now that we have learned quotient groups my next topic is to
do what are called isomorphism theorems. These are all very
important theorems that tell you how to understand groups are
isomorphic and if groups are isomorphic then what implications
do they have, and all this use the notion of quotient groups.
(Refer Slide Time: 00:37)

So the next goal is to study the isomorphism theorems for groups.

So there are three such theorems, in this video I will do the first
isomorphism theorem, and I will illustrate it with one example,
and in future videos we’ll do the other two isomorphism theorems.
So first isomorphism theorem says the following. So there are
three such isomorphism theorems, we will do the first today, first
isomorphism theorem is the following. Let phi from G from G1 be
a group homomorphism, the first isomorphism says that in this
situation we have the following: then we have a group
isomorphism from G/ker phi to image of phi. So that’s what first
isomorphism says, it’s a good time to recall what is an
isomorphism.
An isomorphism is any group homomorphism which is a bijective
set map, so isomorphism allows you to think of these two groups
as essentially the same groups, if you have a group

179
homomorphism kernel in this in order to even make sense
remember I am saying this is a group isomorphism, so these two
are groups in particular, image phi is a group, that we know
because it is subgroup of G prime, but G mod kernel phi also a
group because in general the quotient is only defined when you
have a normal subgroup, but remember kernel phi a normal
subgroup of, it is a normal subgroup. Hence we can talk about G
mod kernel phi as a group, so G mod kernel phi is isomorphic to
image phi.
This should remind you a problem that we solved in an earlier
video where we have shown that, order of G is equal to product of
order of kernel phi and the order of image phi, that is just a purely
set-theoretic statement, number of elements of (G) and that
required you to deal with a finite group, right because the number
of elements of (G) needs to finite for that statement, is the number
of elements of kernel phi times the number of elements in the
image phi. But this is a very general result, this has nothing to do
with finite groups and here we are only talking about
isomorphisms, and in particular that problem follows from this if
you think about it because this two are isomorphic groups means
the orders are same, and the order of (G1 ) kernel phi is order of
(G) divided by the order of kernel phi.
(Refer Slide Time: 3:51)

So the proof of this theorem, let me give, it is very simple, it is

essentially the problem that we solved, in the problem we have
constructed a bijection, now we simply show that bijection is
actually a group homomorphism, consider the function, I think I
called the group homomorphism phi, so let’s called function (F) I
think this what we have done in that problem also, and the
function is the same, what is the function? You take in element of
this which is of the form of (a ker phi), so I think is probably
better if I use letter like before so I am going re-write this as G/N

180
to H1 and what is this? This sends (aN) to phi(a), from a problem
earlier we know that f is bijective, in that problem we exactly
shown it is bijective .
It is certainly onto and if two cosets map to the same element that
means those cosets are same. So we only need now to show that
(F) is a homomorphism, because a bijective homomorphism is an
isomorphism, so we only need to show that (F) is a
homomorphism. What does this mean? Now in other words we
have show that f(aNbN) is want to check f(aN) f(bN), you want to
check this, so let us check both sides, what is f (aNbN) this is
same as f(abN) because that’s the product in the quotient group
(aN) (bN) is (abN) and what is this? f(aN) is phi (a) and f (bN) is
phi (b) right, and what is f(abN) you take any coset and map you
take the element here and map it to phi of that element, so it is f
(abN) will be phi of (ab). Now is this true? phi (a) phi (b) is phi
(ab)? Of course it is true, this is true because phi is a
homomorphism, so we have and hence (f) is a homomorphism. So
we have checked it (f) is a homomorphism. So the proof is done.
(Refer Slide Time: 6:51)

So f is a bijective, bijective comes from before, that we have

checked before and we checked today that it is a homomorphism,
it is a bijective homomorphism so (f) is an isomorphism that’s it,
so (f) is isomorphism and hence we have G/kernel phi, this is my
notation this symbol here arrow with symbol here , to indicate that
we have an isomorphism, that’s the proof of the theorem, first
isomorphism theorem is proved. So its says that if you have any
group of homomorphism, there is an isomorphism between the
quotient group G mod kernel phi and the image group.
So why is this useful? So as I said first isomorphism theorem and
more generally all isomorphism theorems are very useful in
determining properties you know observing new facts about

181
groups. So let us explore this in one example, so I am going to
recall for you, what we have done in a previous video. We recall
that if nZ is a subgroup of Z, actually what we should write is for
any n in Z, the quotient group Z/n Z is cyclic. This I did in one of
the pervious videos, where we talked about quotient groups, we
explicitly calculated quotient groups for subgroups of Z, so in
particular Z/nZ we concluded was cyclic, I left the last part for
you to check an as exercise.
Now we will now show a sort of converse for this, we will now
show that any cyclic group is isomorphic Z/nZ for some n in Z,
this is the very remarkable statement right, it is okay to say that
Z/nZ is cyclic, that is just a statement about subgroups of integers
and quotient groups, now we are saying that you take an arbitrary
subgroup cyclic group, this cyclic group can be anything, nothing
to do with Z to begin with, however it is isomorphic to Z/nZ for a
suitable integer.
(Refer Slide Time: 9:54)

So let me prove this theorem: if G is a cyclic group, then there

exists an integer N such that, this symbol here, my notation I
should remind you, we write G with this symbol to mean G is
isomorphic to (G prime), that’s my notation, so I am saying that
any cyclic group is isomorphic to this.
So this is very, it is to be thought of as a structural theorem for
cyclic groups, any cyclic group is of this form, if you understand
this group then you understand all cyclic groups right, that’s the
meaning, so what is the proof? Proof is an immediate application
of the first isomorphism theorem, so let’s proceed this way.
Let (a) be a generator of G, remember G is cyclic group, that
means it is generated by a single element right, so G is so this is
what it means, powers of (a) cover all of G, this what it means
because such a thing exists because G is cyclic, we are assuming

182
that G is cyclic group, so there is an element (a) such that its
powers span or cover all of G, so you have (a) power N equals, the
set (a) power N, as N various over integers is equal to G.
Now consider the function phi from Z to G, what is this function?
It takes N to a power N, this is a function that we have studied in
the past, in a previous video, so we have earlier that phi is a
homomorphism.
(Refer Slide Time: 12:58)
A quick recall, remember we are defining phi (N) is (a) power (N)
we want phi of (N + M) this is by definition (a) power N+M, this
is same as aN aM that is, if you write (a) product with (a) itself N
times and M times you pool them and you get N+M times and
that is phi (N) phi(M), this is not difficult at all right, so phi is a
group homomorphism.
So not only that so phi is a group homomorphism, so phi from Z
to G is a group homomorphism, and this is the first point, and of
course phi is onto. See this is the meaning of, this is because G is
cyclic, G is generated by, the point is phi maps an element in
integer N to (a) power N, in general the set of elements (a) power
N as N varies is not all of G, that happens only if G is generated
by (a), so G is generated by (a), so phi is onto, because every
element of G is form of (a) power N. Now what is kernel? Is that
clear, phi is onto because G is generated by a, every element of G
is a power an integer, so phi is onto. Kernel phi whatever it is, is a
subgroup Z, right in general kernel is always subgroup of the
group, domain group, so suppose and what are the subgroups of
Z? We know that subgroups of Z are, they are the form, I should
use some new letter may be but hopefully it’s not confusing to use
N again, though we have used N to denote the arbitrary integer
right, there are of the form NZ. So suppose now coming back to
specific kernel phi, kernel phi is NZ, so let’s summarize what we
have.

183
(Refer Slide Time: 15:49)

So we have a homomorphism, group homomorphism phi from Z

to G such that phi is onto and kernel phi is NZ. Now let’s think
about what first isomorphism theorem says, what does first
isomorphism say, what does it say? First isomorphism theorem
says, it says, so let’s see what it is, first isomorphism theorem said
if phi is a group homomorphism from G to G1 then we have an
isomorphism from G mod kernel to image phi.
So we have done now, just apply that to our, so we have Z mod
kernel is isomorphic to image phi, but this kernel is NZ, so Z/NZ
its isomorphic to, because phi is onto, image phi is all of that, so
phi is onto means, image phi is G. So Z/NZ is equal to isomorphic
to G, which is exactly the theorem that I wanted to prove. So we
wanted to prove that any group, any cyclic group, if G is a cyclic
group there exists an integer N such that G is isomorphic to
Z/NZ. We have shown that right, so this completes the proof of the
theorem.
So first isomorphism theorem is a very simple statement really
because its proof is clear, but it’s a very powerful result, it comes
up all the time, it’s a very simple observation but it’s always
repeatedly used in various situations and using it we have proved a
remarkable theorem that any cyclic groups is isomorphic to Z/NZ,
why do I say it is remarkable? Because G is an arbitrary cyclic
group, G could be anything, G is something that we do not know
what it is, however we have identified with a very specific well-
known easily understood group Z/NZ. So this is to be thought as,
the conclusion to be drawn from all this is the following.
Using isomorphism theorems, we can essentially isolate the
important properties of the group, cyclic groups, and think as them
as Z/NZ, there is nothing more.
(Refer Slide Time: 19:18)

184
So let me give you one example to illustrate this. So let G be the
group of 4th roots of unity, what are they? This is a group that we
have seen before, this is (1,I) I being a square root of minus 1
right, complex square root of minus (1) and minus (I). We have
seen before that G is cyclic, and it is of order 4, so actually by the
theorem G is isomorphic to Z/nZ, so Z/4Z.
So I should have been more precise in my statement of the
theorem, I wrote here G is a cyclic group then there exists the
integer N such that G is isomorphic Z/NZ, but what is N actually?
N cannot be arbitrary number, N has to be related to G, here N is
the order of G right, so what we can conclude is why it is equal,
what is the point here, so why is N equal to order of G, because if
go you back and see, kernel will be NZ so, what I said is kernel is
a subgroup of Z , so it is the NZ, but what is that N, it is the least
positive integer such that (a) power N is identity, which is exactly
the order, so N is the order of (a) which is, because (a) generate G
order of (a) is cardinality of G, so in this case G is isomorphic to
Z/4Z.
So to be precise what we are doing is consider the map from Z to
G we send any integer to I power A, so the kernel of this is
precisely 4Z, because kernel is the set of integers such as I power
A is identity, in this case one, so I power 4 is one, I power 8 is one,
I power minus 4 is 1 and so on.
So the theorem really written in a more precise fashion is if G is a
cyclic group of order N then G is isomorphic to Z/NZ.
So let me stop the video here, we have seen a very important
theorem here, called the first isomorphism theorem, and if you go
back and see the statement and its proof, it’s not difficult at all,
however it is always used, it is an important property of group
isomorphisms that comes up a lot, so and we have seen one
example of it where we have concluded that any cyclic group of

185
order N is isomorphic to Z/NZ. So let me stop the video here, in
the next videos we will study the second and third isomorphism
theorems. Thank you.

186
 NPTEL
ONLINE COURSE
Introduction to Abstract
Group Theory
Module 04
Lecture 22-
Examples and Second isomorphism Theorem
PROF- KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE
In the previous video we looked at the first isomorphism theorem
and some examples of it, so I am going to before continuing today,
I will recall for you what is the first isomorphism theorem.
(Refer Slide Time: 00:32)

First isomorphism theorem, it says that, let phi from G to G prime

be a group homomorphism, in this case, then we have an
isomorphism from G mod kernel phi to image of phi. So this is the
statement of first isomorphism there, it tells that if you go modulo
the kernel, you get a group that is isomorphic to the image group,
remember kernel is always normal subgroup, so G mod kernel phi
has a group structure, it is isomorphic to the image of the group,
image of the homomorphism.
So image of phi is a subgroup of G prime okay, and we saw one
example on this, we saw that if G is a cyclic group of order N, G is
a cyclic group containing N elements, then we saw that G is
isomorphic it to Z mod NZ, so any cyclic group of the form Z mod
NZ, and this was an immediate corollary of first isomorphism
theorem, because we look at the homomorphism from the integers
to the group G sending 1 to the generator of the group G, so then

187
kernel will be precisely NZ, it is an onto map, so there will be
isomorphism like this.
Let us look at another example of first isomorphism theorem, so
recall that C* is the group of nonzero complex numbers with
multiplication, these form a group under multiplication, and let’s
denote R+ to be positive real numbers also with the multiplication.
So we saw in previous videos, these are groups under
multiplication. Consider the map, consider the function, let’s say
phi from C star to R+ which sends a complex number to its
absolute value. So if you recall, absolute value is simply this
function, A+IB is a complex number, so A and B are real numbers,
I is a √ -1, you send it to A2 + B2, that is this function, absolute
value of A+IB is A2+B2 , so in order to see what the first
isomorphism theorem has to say about this, let us look at the
kernel of this map, so what is kernel phi? By definition complex
numbers with absolute value 1 okay,
So if you represent complex numbers in the plane, so this the R
real axis, this is the imaginary axis right, so this is the kernel phi,
it is simply the unit circle, it denoted by S1, this is the set of
elements, complex numbers, which have absolute value 1, the unit
circle around the origin, that is the kernel phi. What is the image
of phi? So image, remember phi is a function from C star to R+
positive real numbers, if R is a positive real number, then absolute
value of R is simply R, so phi is actually onto, in other words, the
image of phi is all of R+.
So what does the first isomorphism, now say? The first
isomorphism theorem gives C* mod S1 is isomorphic to R+ okay,
so the first isomorphism says C*/S1 is isomorphic to R+ okay, you
should spend some time to think about this example, and this will
hopefully help you with understanding quotient groups, so this as
groups, C*/S1 is the set of left cosets right, what are left cosets?
These are left cosets of S1 is C*, these are the form {Z S1, Z € C *
}. That is simply the set, right, and on this there is a group

188
structure, remember of course that C star is an abelian group, if
every subgroup is normal and in any case S1 is a kernel of a group
homomorphism, normal, so think of this, the groups C*/S1, as it is
the left cosets of S1, but think of them as circles of varying radii
okay, so you have radius 1 circle.
That is S1, radius 2 circle, radius 3 circle, radius 1.3 circle, so and
if you think about this, if you take all such circles radius will be a
positive real number, you take all circles of positive radius, the
cosets can be represented by those positive reals and hence we
have an isomorphism like this, because if you have all cosets, they
are represented by simply the radius, cosets are simply circles of
different radius, so they are represented by positive real numbers,
so that is a slightly intuitive explanation of this isomorphism,
okay.
One more example let us do, this is something that you have seen
before.
(Refer Slide time: 07:42)

Remember GLn (R) is the group of invertible n by n real matrices.

Consider the map from GLn (R) to R*, nonzero reals, the
determinant map. So a matrix A goes to determinant of (A), since
A is invertible, determinant of A is nonzero, it is a real number, so
it is in R * it is a certainly group homomorphism, that we have
remarked earlier. det is clearly onto also, because if you take any
real number, nonzero real number, you can construct a matrix with
that determinant, for an example you can just take a diagonal
matrices with that number being one of the entries and all the
diagonal entries being 1.
(Refer Slide Time: 09:25)

So this is easily seen to be onto. What is the kernel of det? This is

the group of matrices in GLn (R) which have determinant 1, we

189
called this SLn (R), in an earlier video. So the first isomorphism
theorem gives GLn (R) /SLn(R) os isomorphic to R star, okay, so
this is also a consequence of first isomorphism theorem.
So in general if you want to prove statements about quotients
groups, or solve problems involving quotient group and
isomorphisms involving quotient groups, immediately you should
think of first isomorphism theorem, and the other two
isomorphism theorems I will do later in this video. One remark I
will make before I continue with second isomorphism theorem.
(Refer Slide Time: 09:48)

Let us say that G to G prime is a group homomorphism. Then

G/kernel of phi is isomorphic to a subgroup, this is exactly
statement of, this is an immediate consequence of the first
isomorphism theorem, because first isomorphism theorem says G/
kernel phi is isomorphic to the image of phi, which is certainly a
subgroup of G prime, so G/kernel phi is isomorphic to a subgroup
of G1 and hence G/kernel phi can be identified with a subgroup of
G1.

So it is for all practical purposes, we will take it as subgroup of G 1

so we usually think of G/kernel phi as a subgroup of G 1, so
sometimes is useful to think of G mod kernel of phi as a subgroup
of G prime. In particular if kernel phi ={e}, which is to say phi is
1-1, then G can be thought of as a subgroup of G 1, if G to G 1 is (1-
1) group of isomorphism that mean it is injective, G naturally can
be thought of as a subgroup, because its image is isomorphic to G,
by the first isomorphism theorem. We can replace by G by its
image and think of G itself as a subgroup of G1 okay.

Now that we have looked various examples and a remark about

first isomorphism theorem, let us go ahead and do the remaining

190
two isomorphism theorems. So the second isomorphism theorem
says the following.
(Refer Slide Time: 12:18)

Let G be a group, okay, I am going to consider two subgroups of

G, okay let H and N be two subgroups of G, assume that N is
normal in G okay, I am not assuming anything about H, other than
that it is a subgroup but I am assuming that N is actually a normal
subgroup of G, then the second isomorphism theorem says that H
intersection N, so it says several things, let me may be I should
write the separate statements.

Then H intersection N is a normal subgroup of H, the product set

HN is a subgroup of G and N is a normal subgroup of HN, and
finally in (3) we have H modulo H intersection N, by (1) H
intersection N is a normal subgroup of H, so this is a group by (2)
N is a normal subgroup of HN, so I can consider this okay, So H
mod H intersection N is same as HN mod N.
Whenever you see in books statements about second isomorphism
theorem, this is what we usually refer to as second isomorphism
theorem, the statements (1) and (2) I wrote here are the simply
needed to makes sense of the third statements right, because H
intersection N needs to be normal subgroup of H, to talk about the
left hand side group, similarly HN needs to be group, and N needs
to be a normal subgroup of HN, in order to makes an sense of
right hand side as a group, so this is again group isomorphism.

I have been using this symbol to always denote group

isomorphism, okay. Let us prove this quickly, it is not difficult at
all. Okay statement (1) we know that H intersection N, so what is
the meaning, so certainly H intersection N is contained in H, it is a
subgroup obviously, because H intersection N being an

191
intersections of groups it is a subgroup this is a very exercise
because identity is in H and it is in N, so it is in H intersection and
N, if the two thing are in intersections their both in H so the
product in H they are both in N, so their product is in N. So their
product is also in the intersection, so this is a very exercise.

In order to check that it is a normal subgroup what do we have to

check? So let us choose h in capital H and n in H intersections N
okay, so this we need to check that we need to check that h n h
inverse is in H intersection N right, because this is the definition
of the normal subgroup.

So recall we say that H is normal in G, definition, if g is in G h is

in H then g h g inverse is in H. So you take something from the
bigger group and something from the smaller group and consider
this element, it must be in the smaller group. Here we are
interested in proving that H intersection N is normal in H so we
want to take something from H something from the smaller group
and consider this element and show that it is in the smaller group
okay. First of all note that h n h inverse is in N.

Because h is in G and n is in N and N is normal in G, right . So

this is because N is normal in G, that is hypothesis so we are given
that N is a normal subgroup of G so you take something in H and
it obviously in G because H is a subgroup of G, so small h is in the
group G, small n is in group N and the element h n h inverse will
be in N. On the other hand, h n h inverse is also in H, because h is
in H, n is in H intersection N which is contained in H, so h is in H,
n is in H and H is a subgroup. Note that because H is a subgroup
and small h is in it and small n is in it, small h, small n small h
inverse will be in H, so these two thing together imply that h n h
inverse is in H intersection N. So as we needed so (1) is proved
because H intersection N is a normal subgroup of capital H.

(Refer Slide Time: 19:01)

192
(Refer Slide Time: 19:05)

Let us prove (2). (2) remember says two things, two statement are
made in (2), it says that, it say that, first of all it says that capital H
times capital N is a subgroup of G and that capital N is a normal
subgroup of capital HN.

(Refer Slide Time: 19:24)

So first of all what is capital HN? This I defined in an earlier

video, HN and in general for any two subsets of a group the
product is simply this set, you take elements from the first set and
elements from the second set and multiply in that order, so you
take small h from capital H small n from capital N and you
multiply them and you vary small h and small n okay so why is
this, so this is certainly a subset of G.

So why is this a subgroup of G? So it is important to note here that

if you take two subsets their product is certainly not in general a
subgroup. In fact even you if take two subgroups their product is
in general not a subgroup. However if one of them happens to be
normal it is a subgroup okay let us check one by one the required
properties. E belong to H and E belong to N certainly that is
because both of them are subgroups.

193
So E times E which is in HN, remember in order to be subgroup
capital HN must contain the identity element, it must be closed
under multiplication, it must be closed under taking inverses. We
have checked that it closed it contains the identity. Let us take now
two elements of the set HN, so H1 N1, H2 N2 let say these are
two elements are HN, we want to prove H1 N1 times H2 N2 is in
capital HN again.

But here we have to use normality of capital N, so note that N is

normal G, this implies (H2 inverse N1 H2) is in N because H2 is
an element of the group G and N1 is in capital N, so this element
is in capital N, so this elements is in capital N, so say (H2 inverse
N1 H 2) is equal to some N3, it is equal to some element of capital
N, call it N3. This gives me (N1 H2) is (H2 N3). I am simply
multiplying by H2 to the left on both sides. Now what is (H1 N1
H2 N2)? This is equal to, because I do not, remember because of
associativity I do not to be careful about putting brackets. N1 H2
is equal to H2 N3, by that calculations so this is H2 N3 N2. This
can be written as (H1H2 N3 N2), now H1 H2 are in capital H so
this is in capital H product is also in capital H similarly this is in
capital N, so this is in HN. So we have said that two things in
capital HN means their product is also in HN.

Finally, how do you show inverses? It is the same idea, if you

have HN in capital HN, what is HN inverse? This is N inverse H
inverse okay, by what we have shown here we have shown that
(N1 H2) is (H2 N3), so here what we have really shown is that
anything in N times anything in H is you can switch the two you
can replace you can bring the H2 to the front but on the right now
there will be some element of N, so this by the same calculation as

194
above equal to H inverse times some N1, by the same calculation
as above, okay and this is in capital HN. So the point that I want to
highlight here and this the point that we used above.

If N is in normal in capital G then for g in G, gN is equal to Ng.

See elements of N do not necessarily commute with a fixed
element but as a coset they commute, this is something that we
have seen before also but this is exactly the calculation that we
have done here. So something in N times G, G times something
else in N so this is the point that I would like you to remember. So
this is what we have used, so HN is a subgroup of G so this
certainly proves the a part of the second statement of the theorem
and also N is E times N is contained in HN right.

Because HN is product of elements of capital H and elements of

capital N, E in an element of H, so you have this. So N is a
subgroup okay. To finish the proof of statement 2 of the theorem
we want to show that N is actually normal in capital HN but that is
obvious because N is certainly normal in HN, because N is normal
in even a bigger group, namely G. See what is the picture now we
have N sitting inside HN sitting inside G. N is normal in capital G
that means you take something in capital G take something in
capital N and do that operation GN G inverse, it is in capital N for
G in G and N in N, this is certainly going to hold if you take
something in capital HN and do this same operation so because
HN is already a subgroup of capital G, so if you take something in
capital HN it is in G.

So this condition holds, so if you have a subgroup which is normal

in the ambient group G then it will be normal in any other

195
subgroup that contains it, so this proves (2), right. (2) is
completely proved because (2) is the statement that HN is a
subgroup of G which we have showed and we also showed that N
is a normal subgroup of HN.

So now let us finally prove the main part of the theorem, which is
the isomorphism part of the isomorphism theorem. Now we want
to show that, here we will show, so just to recall H mod H
intersection N is isomorphic to HN mod N, this is what we want to
show. So let consider in order to this, let us consider the map from
H to HN mod N, let us call this phi. What is this map? phi of a
small h is by definition hN, okay so this is the map. So I want to
first of all understand what this really means, remember that, so if
you take, what is HN?

So note what is HN mod N? What is HN mod N? These are all

elements all cosets, note HN mod N are the left cosets of N in
capital HN, so these are elements of the form AN where A is in
capital HN, so in general this is the quotient, quotient group, AN
where A is in HN, but capital H is contained in HN because of the,
I mean this is similar to in previous part when I said N is
contained in HN. H Is contained in HN because identity is
contained in N.

So we can talk of HN, so hN belongs to HN mod N, if h belongs

to H, so this is the explanation for this, I am sending h to hN and
keeping in mind that small h is also an element of capital HN. So
we want to claim that phi is onto and kernel of phi is H
intersection N. Once we prove the claim the theorem will follow

196
immediately from the first isomorphism theorem, so what is the
proof of the claim? So why is phi onto? In order to prove phi is
onto we want to show that anything in the right hand side is an
image of something.

So let AN be element of HN mod N, so A is inside HN, right so A

is an element of HN right, it a left coset of capital N in capital HN,
so A is an element of HN. So write A as small h times small n,
where of course small h is in capital H small n is in capital N.

Now let us look at what is AN, AN is HN N. So this can be written

as HNN but what is NN because N is in capital N, NN is just HN
so AN is equal to small h times capital N, but this remember is phi
H. So we started in arbitrary element of, started with an arbitrary
element of HN mod N and showed that it is the image of
something in capital H, so this implies phi is onto right, so phi is
onto, this much is okay.

What is the kernel of phi? kernel of phi is all elements of group H

such that HN phi so the definition is phi of H must be in the
identity element. What is the identity element of HN mod N, HN
mod N is the quotient group, identity element is simply E times N,
so this is the kernel, but this is equal to all elements small h such
that hN equals N, EN is equal to N, so I am going to write it as
things that map to N are the elements in the kernel. But what does
hN equal to N mean? So if hN equal to N, I claim this is true if
and only if h is in N, right, this we have seen when we discussed
cosets, hN is equal to N if and only if h is in N, certainly so
because if h is in N then hN and N have something in common.
So they must be identical, any two cosets are either identical or are

197
disjoint, so if h is in N, h is N as well as h is in HN because h is h
times identity, so they are identical cosets. On the other hand, if
hN is equal to N, then certainly h will be in N because h is in HN,
okay so this implies kernel phi, so these are elements I am going
to write it simply here, h in H such that h is in N. But what is this
set, this set is nothing but things that are in H well as in N, so
kernel of phi is simply H intersection N, so the claim is proved.

The claim, claim that phi is onto the kernel phi is H intersection N.
Now the first isomorphism theorem says, so the phi is a map from
H to HN mod N, phi is onto, kernel phi is H intersection N. So by
the first isomorphism theorem, first isomorphism theorem applies
to the any group of homomorphism, it says that G mod kernel is
isomorphism to image, so in this case image is all of this HN mod
HN, and kernel is H mod H intersection N. So H quotiented by H
intersection N is isomorphic to H mod the kernel which is H
intersection N is isomorphic to image which is all of HN and mod
N. Remember this is the theorem that we wanted to prove. The
second isomorphism says H mod H intersection N is isomorphic
to HN mod N and we have completed the proof of the theorem
okay.

This is the, this finishes the proof, okay, let me stop the video
here, in this video we have looked at some more examples of the
first isomorphism theorem and we have stated and proved the
second isomorphism theorem. In the next video I will do the third
isomorphism theorem, thank you.

Online Editing and Post Production

Karthik

198
 Ravichandran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh Kumar
Web Studio Team
Anitha
Bharathi
Clifford
Deepthi
Dhivya
Divya
Gayathri
Gokulsekar
Halid
Heamvathy
Jagadeeshwaran
Jayanthi
Kamala
Lakshmipriya
Libin
Madhu
Maria Neeta
Mohana

199
 Mohana Sundari
Muralikrishnan
Nivetha
Parkavi
Poonkuzhale
Poornika
Premkumar
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirley
Subash
Suriyaprakash
Vinothini
Executive producer
Kannan Krishnamurty
NPTEL Co-ordinators
Prof.Andrew Thangaraj
Prof. Parthap Haridoss
IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resource Development

200
Government of India
Copyright Reserved

201
 NPTEL

NPTEL ONLINE COURSE

Introduction to Abstract

Group Theory

Module 04

Lecture 23-“Third isomorphism theorem”

PROF.KRISHNA HANUMANTHU

CHENNAI MATHEMATICAL INSTITUTE

So far we have looked at first and second isomorphism theorems,

in this video let us start with the third isomorphism theorem .

(Refer Slide Time: 00:31)

Recall that isomorphism theorems are statements about

isomorphisms among quotient groups, okay, the third
isomorphism we will say similar statement to the first and second,
so let us see what it says. Let G be a group okay let H, N be two
normal subgroups of G, okay, in the second isomorphism theorem
we worked with two groups one normal, and the other arbitrary, in
this case we take both to be normal H and N are both normal
subgroups of G. Then we have two statements, actually I should
write let H and let H and N be normal groups of G such that N is
contained in H and H is contained G of course. So I am taking two
subgroups one containing the other, N is contained in H and both

202
are normal, N and H are both normal in G, then we have two
statements.

Then G then H/N, because N is an subgroup of H I can consider

H/N, remember N is a normal subgroup of G so it certainly a
normal subgroup of H, I want to say H/N is a normal subgroup of
G/N. So we will explain this, so H/N first of all I have to think of
it as a subgroup of G mod N and then I want to further say it is a
normal subgroup of G mod N. And the second statement which is
actually what we called third isomorphism theorem is that G/H is
isomorphic to G/N mod H/N, okay. So this migHt seem ugly
here but the point is H/N is a normal subgroup of G/N by the
statement one.

So I can consider the quotient group G mod N modulo H mod N,

that is same as G/H. This is a very useful theorem actually, third
isomorphism theorem, along with the first isomorphism theorem,
they both get used a lot. Here think of this as a sort of canceling,
G/N mod H/N, is like almost like when you write ratios of
numbers like integers you can cancel, something like that you can
do here, that is why this is very useful to write it like this, G/N
mod H/N is isomorphic to G/H, that canceling is valid is what the
third isomorphism theorem says.

So first of all let us tackle (1). We want to show that H/N is

normal subgroup of G/N, in order to do that we first want even
think of H/N as a subgroup of G/N, before we even ask if it is a
normal subgroup, is it a group subgroup of G/N? Note that we
have H is a subgroup of G, so we have an inclusion H in G rigHt.

So this inclusion is simply sending small H to small H. We also

have a natural homomorphism G to G/N, so let us put these two
things together, we have H to G which is a natural inclusion, so in
particular it is an injective map, it is a 1-1 map and then we have
the natural surjective map from G to G/N, remember when I talked

203
about the quotient groups I always defined this natural map that
exists always g goes to gN, this is a group homomorphism that is
the second map G to G/ N.

(Refer Slide Time: 5:25)

So considerer the composition, this composition, what is a

composition? H to G/N rigHt but it is actually going via G, so first
you send it to G then you send it to the G/N, so this is really
nothing but small h going to hN because under this composition
here you see that under the first map you take a small h it goes to
small h under the first map, in the second map takes small h to
small hN so under the composition small h goes to small hN okay.

(Refer Slide Time: 07:34)

What is the kernel of this? Kernel of this map is {h in H| hN=N}.

Remember N is the identity element of G/N, identity coset is the
identity element, but as before this is {H in H/H in N} in the
second isomorphism theorem also this came up, so it is h in H but
it is also in N, this of course is H intersection N, but in the new
hypothesis remember N is also a subgroup of H, so this is just N,
so kernel of this map is N.

So by the first isomorphism theorem, H/N is isomorphic to the

image of this map, I don’t care what it is, it is a subgroup of G/N
is all care about, is isomorphic to a subgroup of G/N and as I have
remarked in a previous video after I did the first isomorphism
theorem, if you have a situation like this, the technically statement
is H/N is isomorphic to a subgroup of G/N, but we can think of as
a subgroup of G/N.

204
So we think of H/N as a subgroup of G/N, we are claiming in the
first part of the theorem that H/N is a normal subgroup of G/N. I
have justified that it is a subgroup of G/N, now we have to show
that it is a normal subgroup. What is normal subgroup? So we
have H/N inside G/N, normality means what? You take something
from the bigger group something from the smaller group and
consider the element by taking inverse and product.

So what is an element of G/N it is of the form gN, what is an

element of H/N, it is an element of this form hN, because these are
left cosets of N in H, so g is in G, and h is in H. And what do we
consider? We will consider this (gN)(hN)(gN) inverse. Lets
understand this element, what is this, this is really a operation of
cosets, but we have worked with cosets now in several videos, this
hopefully it will be clear to you, it is easy to see that this is (gN)
(hN), remember in the quotient group G/N the inverse of the coset
gN is simply given by g inverse N, this is also something we have
shown, while proving that G/N is a group when N is a normal
subgroup in G, gN whole inverse is g inverse N okay.

(Refer Slide Time: 13:12)

So now let’s write it like this, so we have (gN) (hN) (g inverse N),
so we have this because of associativity we can rearrange our
brackets, so I am going to write basically I am getting rid of this
bracket and write it like this, let us look at this part N, g inverse. In
some previous video I commented that if N is normal so note that
if N is normal in G then gN =Ng for all g in G, rigHt. So this
means that Ng inverse would be equal to g inverse N, so gN, H, g
inverse NN, so I am switching Ng inverse writing it as g inverse N
but this gN h h inverse and N N is N, this is a property of left
cosets, products really of subgroups.

205
(Refer Slide Time: 12:14)

So this is N, but now I can, using the same property that gN = Ng,
pull h and g on this side, so this will be gH N g inverse N, which
is same as I am interchanging N and h here, now I will do Ng
inverse. So gH, g inverse NN which is of course again gH g
inverse N okay, so thougH this seems somewhat magical here
some of you migHt worried about whether all this is valid or not,
you have to convince yourself that this is valid because all I am
doing here is multiplying subsets of group and multiplying subsets
of a group follows the same rules as multiplying elements of that
group and we are using here the properties of normal subgroups, if
N is a subgroup then NN equals N and everything else is correct,
so all this is a valid operation okay.

(Refer Slide Time: 13:10)

Let us now take stock, where are we, gN hN g inverse N is equal

to gHg inverse N. Now I want to use the normality of H in G, gH
g inverse belongs to H rigHt. Because g is in capital G h is in
capital H so gH, g inverse is in, g is in G h is in H, so this is in H,
so gH, g inverse N is in H mod N, because this is in H this is a left
coset of N in H, which is by definition an element of H mod N. So
now we have proved that H mod N is a normal subgroup of G
mod N, we have started with an arbitrary element of G mod N ,
and an arbitrary element of H mod N and looked at gN hN gN
inverse and concluded that it is in H/N, which is the definition of
H mod N being normal in G mod N. So this is (1). (1) is exactly
that statement and it is proved now. So let us prove 2 now.
(Refer Slide Time: 14:29)

206
2 is the statement that G mod H is isomorphic to, we will show G
mod N, G mod H is isomorphic to G mod N mod H mod N. So
how do we show this? So again whenever you have a statement
that you have to prove which involves proving some quotient
groups are isomorphic to each other, think of constructing
appropriate group homomorphisms, and invoke first isomorphism
theorem.
As we have seen in the proof of the second isomorphism theorem,
it is really an application of the first isomorphism theorem. And
here also it will be an application of the first isomorphism
theorem. The map to consider here is the following. Let me
consider this map from G mod N to H mod N. What is this map? I
will take a coset, element of G mod N which is a coset of capital N
in capital G.
So I will take a small g in capital G, so this is not the map to H
mod N, this is the map to G mod H. So take a small g in capital G,
consider the coset gN and I will map it to gH. So this is the map
gN goes to gH, certainly gH is an element of G mod H because it
is a coset of capital H in capital G. So φ must be well-defined in
order to be a group homomorphism. Before that you have to check
well definedness, why do we need to check well definedness here,
because we can have gN equal to g prime N and this can happen
for two different elements of the group G even if they are different
gN could be equal to g prime N.
Suppose this happens, we want to check that φ gN equal φ g prime
N. otherwise it is not well defined, you have one element which
has two different representations and using the representations if
you send it to different elements, it is certainly not a well defined
map, so we want to check that gH is equal to g prime H. But if gN
is equal to g prime N remember this means that, g inverse times g
prime is in N.

207
This is because you can multiply by g inverse so N will be equal
to g inverse g prime N, and that means g inverse g prime will be in
capital N. This kind of thing we have seen before, if g inverse g
prime is in N, then g inverse g prime is in H, because N is
contained in H, by hypothesis N is contained in H. So we have
this, but if g inverse g prime is in H, gH equals g prime H.So well
definedness is okay.
If you have two representations of the same coset, they will also
give you the same coset in G mod H. So if gN equals g prime N,
gH equals g prime H. So this is okay, and φ is a group
homomorphism, this is easy because what is φ of gN times g
prime N. This φ of g g prime N, this is by definition g, g prime H.
And this is equal to gH g prime H, and this is φ of gN, but so φ is
a group homomorphism. I am checking every detail here, though it
is not difficult to check them. What is the image of φ,
(Refer Slide Time: 18:48)
I claim φ is onto. Why? φ is an map from G mod N to G mod H.
So let what is an element of G mod H, an arbitrary element of G
mod H is a left coset, it is gH. Then φ of gN is gH, so I take the
same element small g , gN maps to gH. So φ is onto, this is trivial,
you take a coset is is certainly the same element times N and maps
to it. So it is onto. What is kernel of φ? This is what we have to
check now, what is kernel of φ?

So kernel of φ is by definition all elements of G mod N, it is all

elements of G mod N, which are denoted by g capital N and
remember these are cosets, all elements of G mod N, such that φ
of gN is equal to the identity coset.

Identity element of G mod H is H. So this is equal to that, but this

is gN in G mod N, gN in G mod N, such that gH equals to H.
Remember φ of gN is by definition gH. So φ of gN is equal to H

208
means, gH equal to H. That means gN is contained in G mod N
such that g is in H. So these are cosets of the form gN in G mod N
such that the small g is in capital H.
But what is this, if you think for a second about this, these are left
cosets of capital N in capital G, which are also left cosets of
capital N in capital H, because this g must be in capital H. So this
are just in H/N. H mod N remember is hN, h is in H. H mod N is
the cosets of capital N in capital H, so that are of the form hN,
small h in capital H. But this is really what we are doing here, we
are taking all g in H and taking G mod N, gN sorry. All gN instead
of calling it small h I am calling it small g here.
So if we are taking small g in H and taking gN, so this is exactly H
mod N, so kernel of φ is H mod N.

Now what is the situation, φ is a map from G mod N to G mod H,

it is a group homomorphism, onto, kernel equal to H mod N. Now
what does the first isomorphism theorem say, again it is the final
point is the consequence of first isomorphism theorem, you have a
group homomorphism which is onto and kernel is H mod N.
So you have G mod N which is the domain group moduo the
kernel which is H mod N, I have computed that, is isomorphic to,
remember G to G prime is a group homomorphism means, G mod
kernel is isomorphic to image, so I am applying it to G mod N to
G mod H, it is a group homomorphism, so G mod N modulo
kernel which is H mod N isomorphic to the image which is G mod
H, because it is onto and this is exactly what we want to show, G
mod N mod H mod N, is G mod H.
So we have finished the proof. So this is exactly the isomorphism
that we have claimed and we have proved this. So this completes
the proof of the third isomorphism theorem. So together this first,
second, third isomorphism theorems are very important and as we

209
have seen in the proofs it is clear that the crucial observation came
from the first isomorphism theorem.
Then it is just looking at more and more special cases to get
second and third isomorphism theorems and these are used
frequently in many theorems and in future classes we will see this,
so it is important that you carefully work out the proofs and make
sure that you understand all the details. So I will stop here, in the
next video I will look at some more applications of these quotient
groups. Thank you.
Online Editing and Post Production

Karthick

Ravichandaran

Mohanarangan

Sribalaji

Komathi

VigNesh

Mahesh kumar

Web Studio Team

Anitha

Bharathi

Catherine

Clifford

210
 Deepthi

Dhivya

Divya

Gayathri

Gokulsekar

Halid

Hemavathy

Jagadeeshwaran

Jayanthi

Kamala

Lakshimipriya

Libin

Madhu

Maria Neeta

Mohana

Mohana Sundari

Muralikrishnan

Nivetha

211
 Parkavi

Poonkuzhale

Poornika

Premkumar

Ragavi

Raja

Renuka

Saravanan

Sathya

Shirley

Sorna

Subash

Suriyaprakash

Vinothini

Executive Producer

Kannan Krishnamurty

NPTEL Co-ordinates

Prof. Andrew Thangaraj

212
 Prof. Prathap Haridoss

IIT Madras Production

Funded by

Department of HigHer Education

Ministry of Human Resource Development

Government of India

HYPERLINK "http://www.mptel.ac.in"www.mptel.ac.in

CopyrigHts Reserved

213
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 05
Lecture 24- “Cauchy’s theorem”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE
So in this video I am going to prove an important theorem, a
special case of important theorem called Cauchy’s theorem.
(Refer Slide Time: 0022)

So this is a good application of the concept of quotient groups, so

this is useful to you will understand in proof why is it is useful to
look at quotient groups. So Cauchy theorem says that so this is the
following, so let G be a finite abelian group. So it is a finite group
so meaning it has only finitely many elements and it is an abelian
group meaning any two elements in G commute with each other.
Suppose that a prime number P divides the order of G.

So P is a prime number that divides the order of G, then G has an

element of order P, so then G has elements of order p. So this is
Cauchy's theorem. We will first prove this and then look at why
this is a surprising statement. If G has an element of order p it is a
strong condition on the group, so and it is important to note that P
must be a prime number. So all this we will see after we prove the
theorem and see where we use the prime number hypothesis.

214
Again let us see the statement. G has an element of order p what is
an element of order? Order of an element, recall order of an
element A is the least positive integer d such that A power d is
identity. So there is such an element in the group G. so we are
going to consider the size of the group, so we will prove this in
two cases, we will consider two cases. So the cases are the
following.

Case1: G contains, G does not contain a subgroup H such that 1 is

strictly less than order of H which is strictly less than order G. So
remember G is a finite group so I can talk of order of G, is the
number of elements, so suppose G does not contain a subgroup H
which has this property. That means cardinality of H is strictly
greater than 1 means H contains elements that are different form
identity element. And order of H less than order of G means H is a
proper subgroup.
(Refer Slide Time: 04:02)

If G does contain such an element, so then I will first solve this

problem in, prove the theorem in this case. So choose an element
A in G such that A is different from E, E is the identity element.
Why is there an element A such that A is different from E? Such
an element exists because cardinality of G is greater than 1. Why
is cardinality of G is is greater than 1? Because I am assuming that
a prime number p divides the order of G, so if cardinality or order
of G is 1 certainly no prime number divides it, prime numbers are

215
2, 3, 5, 7 and so on, so order of G is at least 2, that means there is
an element A which is different from E.

Consider the subgroup generated by A. Now what are we

considering in this case? This case we are assuming that that G
does not contain a subgroup H which is non-trivial and proper,
because A is different from E, by hypothesis A is different from E
by choice. So the subgroup generated by A cannot be identity and
hence it has to be all of G, because G does not contain any
subgroup that is non-trivial and proper, if A is the subgroup
generated by A, then it is non-trivial because A is different from E,
so it must not be proper so it is equal to G. On the other hand, so
in particular G is cyclic group. What is the order of, note that if
you have a cyclic group generated by A, order of G is equal to
order of A.

Because, this is because, G is nothing but E, A, A 2 , A power

(order of A)-1, you take powers of A and go all the way up to
(order of A)-1, the next power which is A power order of A will be
E, so this is G, how many elements are here? Order of A many
elements are there, so that is the order of G. So now and what are
the assuming? We are assuming that P, we assume we know, P
divides order of G which is order of A.

So now we have to exhibit an element of order P, so if P divides

order of A, so I am going to, so best way to complete now is
through the following exercise, so I will leave it as separate
exercise and do this later in a separate video or may be in the same

216
video, but it will interfere with the proof of Cauchy’s theorem. So
I don’t want to spend time on this, if G is a cyclic group of order
N and M divides N, then G contains an element of order M.

So this is not a difficult exercise at all and this is a very general, so

M, N are positive integers. And it has nothing to do with
primeness, so G is a cyclic group of order N and M is another
number that divides N, then G contains an order of element M so
this so this is sort of a converse to Lagrange’s theorem, which I
proved earlier, what is Lagrange's theorem? It says that G has a
finite group and we have element its order divides the order of the
group.
I am saying now that if I am further assuming G cyclic and a
number divides the order of group then there is an element of that
order.
(Refer Slide Time: 09:56)

Now the by the exercise, G contains an element of order P,

because G is cyclic, that is what we have concluded, G is a cycle
group, so and P divides the order of the group, so G has an
element of order P, so if the group is assumed to be cyclic, very
strong statement is known, much stronger than Cauchy’s theorem,
it says that if P is a prime number, and P divides the order of the
group, and G is abelian, that is the version of the Cauchy’s
theorem that we are doing now, there is an element of order P. But
the exercise is saying that if G is cyclic, life is much simpler, for
any M dividing the order of the group, there is an element of that
order.

217
So I will do this exercise later, but we are done, assuming that
exercise. So now let us go to case two, so in case two what is the
complement of case 1, case 1 was there is no subgroup of G which
is non-trivial and proper, so case 2 would be there is a subgroup
of G which is non-trivial and proper, so there is a subgroup, H of
G, which is non- trivial and proper, that is H is different from the
trivial group, that is H is different from G also, so we have 1
strictly less than order of H which is strictly less than order of G,
okay.

Now I am going to use induction on G, so what does the theorem

say, theorem says that if G is a finite abelian group and a prime
number divides its order then G contains an element of that order,
so what would be the base case of induction, because I want to
take an order, which is divisible by a prime number, order of G is
two, G is a and e and order of G is divisible by two and order of a
is two, so only prime that divides the order of G is two and there is
an element of order two, this is okay, base case is proved.
(Refer Slide Time: 12:56)

So now I can assume that up to some number all groups of that

order which are abelian and if a prime number divides them, there
is an element of order p. So now we know that order of H is
strictly less than order of G, so suppose that P divides order of H,
this is not necessarily true but suppose it is true. We only assumed
that P divides order G. Suppose that P divides order of H.

218
Then by induction because the order of H is strictly less than order
of G and we are looking, induction hypothesis is that anything
which has order less than the order of G the theorem goes, now H
is an abelian group, G is abelian that implies H is abelian,
abelianness is inherited by subgroups, any two elements of G
commute with each other, so any two elements of H also commute
with each other, so H is abelian, its order of H is G and P divides
order of H, so by induction H contains an element of order P.
(Refer Slide Time: 14:51)

So this is the induction statement, but if H contains an element of

order of P then so does G. That is immediate right, if H contains a
of order P, then a belongs to H, order of a is P, implies a belongs to
G, H is a subgroup of G and order does not change, order of a is
order of a, so G contains the element of order of P, so this is the
easy case, if P divides order of H then we are done.

So, suppose that P does not divide order of (H), this can happen
right, P divides order of (H) or P does not divide order of (H), but
now recall counting formula, that we learned while proving
Lagrange's’s theorem. It says that order of (G), is equal to order of
H times order of (G/H), the index right, now since we have
assumed that G is abelian, every subgroup is normal right, in an
abelian group, G/H is a group and note that order of G/H is the
index of H in G, (G:H). Now if you look carefully at the counting
formula, the order of G is equal to order of H times order of G/H
right, and what is the hypothesis? Hypothesis is that, P divides
order of G.

219
And hence P divides order of H times order of G mod H, and a
property of prime number says that, P divides order of G mod H,
see P divides order of G, P does not divide order of H, so P divides
order of G mod H, this is a property of prime numbers, if P is a
prime number and it divides the product of 2 numbers, it must
divide one of them, and it does not divide H, is our hypothesis so
P divides order of G mod H and, note that order of G mod H, is
equal to order of G mod order of H, which is less than, order of G,
also because H is a nontrivial group, this is because order of G,
order of H is greater than 1, so induction hypothesis is okay, so in
order to make sure that, induction hypothesis is applies to G mod
H, we need to check that G mod H is abelian group, is that clear?

If G is abelian, G mod H is abelian, this is a very easy exercise,

that I will do later in a video, G mod H is abelian, order of G mod
H is less than order of G, and P divides, order of G mod H, that is
what I noted here, so induction hypothesis applies to G mod H. So
I want to emphasise a very important point here that, I need to
assume that G is abelian okay, that is why I state in the original
theorem, that G is abelian.

Because if G is abelian only you can consider G mod H for an

arbitrary subgroup H, otherwise G mod H is not a group, it is just
a set of cosets. So now what does the hypothesis say? Induction
hypothesis so, by the induction hypothesis, G mod H contains an
element of order P, right, because G mod H is an abelian group, its
order is divisible by P, and its order is less than order of G, so it
contains an element of order P, but what are elements of G mod P,

220
they are left cosets of H in G, so say gH in G mod H has order P,
right.

This means gH power P is equal to H, because that is the meaning

of order, gH power P is H and gH is not equal to H. If order of gH
is equal to P implies this, P is the smallest positive number such
that gH power P is identity which is H, in G mod H identity is H,
so gH power 1 because remember P is strictly more that 1 always,
P is a prime number so gH cannot be H and gH power P is H. But
what is gH power P that means g power P H is H, right
(Refer Slide Time: 21:12)

and of course gH is not equal to H, that means g power p is H and

g is not in H this is the meaning when you pull the element back to
G. g is an element of G, it is p-th power is in H and itself it is not
in H okay.
(Refer Slide Time: 21:17)

So g power p is in H and g is not H. Now let us denote by m the

order of H, okay, so I am just giving it a name. Let m the order of
H. Then what we have is by Lagrange’s theorem applied to H and
what is Lagrange’s theorem? It says that if you have a group and
an element in it, order of the element divides order of the group.
(Refer Slide Time: 22:08)

221
Order of g power p divides order of H which I called m, okay. If
order of g power p divides m and this is something I have showed
before, g power p whole power m is equal to e. This is the exercise
that, this came up in several places, if g is an element of group G
and order of g is let’s say a and then g power an is e for all n right,
so if here m is divisible by order of g power p, m is order of g
power p times something.

So g power p power m is identity. This means we have g power

pm is identity. I also remarked numerous times, that usual
exponent rules apply to taking powers in elements of groups. So
this is remember I can always switch the exponents. So, g power p
whole power m is identity means g power m whole power p is
identity. So now can we say, that order of g power m is p, can we
say this?

See not quite, right, because what is order of g power m? This is

this least positive integer such that g power m power p is identity.
We have certainly g power d, least positive integer, d such that g
power m power d is identity. We know that g power m power p is
identity, but what if there is a smaller number, we can certainly
say is that order of g power m divides, okay so this is also an
exercise this is sort of converse to this exercise that I wrote in the
box here.
That also we have repeatedly discussed, if you have an element of
order is 3 only we have that a power will be equal to identity if
you take multiples of 3. Right, so recall if order of g is a and g
power n is identity, then a divides n, using this we conclude that
order of g power m divides p, because g power m power p is
identity, so order of g power m must divide p. Right, this is
discussed in various previous videos.

222
So, now hopefully you remember, otherwise go back and see some
of the videos that talk about this. But now p is a prime number so
we are going to use this fact now, and I think this is the first time
we using that p is a prime number right.
(Refer Slide Time: 26:41)

So p is a prime number, so actually it is not first time, we have

also used it here. To say that if p does not divide order of H if p is
does not divide order of H, it divides order of G mod H, here also
we used. So p is a prime number so only numbers that divide p are
1 and p, so order of g power m is 1 or order of g power m is p.
Right, I have observed here that order of g power m divides p, but
p is a prime number so order of g power m is 1 or order of g
power m is p.

Because only numbers that divide P are 1 and P. But order of g

power g is 1 means what? g power m is identity, an element has
order 1 if and only if it is identity, but if g power m is identity that
means, if g power m is identity, apply Lagrange’s theorem, so now
I will do, so g power m is identity let us say. So I want to say that
this is not possible, why? The reason is this is not possible because
g power m is identity right, suppose g power m is identity, this
implies order of g okay, so if order of g power m which we know
either 1 or P if the order of g power m is 1 then g power m is E if
which I clam is not possible.

So why is that? If g power m is E then certainly gH power m is

g^m H which EH which is H so g power H, g H power m is H, but

223
recall how did we pickup gH, order of gH is P, that is the way we
constructed, we chose, because by induction hypothesis applied to
G/H it contains an element of order P so gH was that element, so
gH power m = identity and order of gH = P that means P divides
m by the same exercise as before if order of g is A and g power N
is E then a divides N.

But P divides m, this is not a good thing right, because then P

divides what is m, m was order of H, so we simply called the m
the order of H, somewhere I wrote that right, we let m be the
order of H but we are assuming the P does not divide order of H,
because if P divides order of H we would not even do all this, if P
divides order of H we are done right away, so does not divide
order of H is our assumption and now we are getting the P divides
order of H so g power m cannot be equal to E, so this is not
possible. So order of g power m is P and we are done because
remember we want to exhibit is the element we are looking for, g
power m is the element we are looking for.

We want to exhibit, this completes the theorem proof of the

theorem, we want to exhibit an element of order P. Just to recap,
so this is, you have to carefully follow the proof that I gave, I am
going to spend 2 minutes recalling the proof how we did this.

The Cauchy’s theorem says that if G is a finite abelian group and

P is a prime number that divides the order of G then G has an
element of order P. The first which is easy is that G has no non-
trivial proper subgroups, then we conclude immediately the G is
cyclic. And by this exercise which I will do later we are done.

224
So the interesting case is, there is a subgroup of H of G which is
non-trivial and proper. Then using abelianness of G, we have that
H is normal which we use later. So we know that order of H is
greater than 1 and less than order of G and we use induction, if p
divides order of H, so by the counting formula, p must divide
either order of H or order of G mod H.

Because P divides order of G and P is a prime number. If P divides

order of H immediately by induction H contains and order P
element so G also contains an order P element, the same element.
Otherwise G mod H contains an order P element. Let us call that
gH, then we played with that element and concluded that an
element that is related to it is the element we are looking for.

Okay, so the last part of the proof is somewhat tricky perhaps, so

you have to carefully read the proof and listen to my proof once
more if needed to make sure that you understand, okay, so this
complete the proof of Cauchy’s theorem.
Online Editing and Post Production

Karthick

Ravichandaran

Mohanarangan

Sribalaji

Komathi

225
 Vignesh

Mahesh kumar

Web Studio Team

Anitha

Bharathi

Catherine

Clifford

Deepthi

Dhivya

Divya

Gayathri

Gokulsekar

Halid

Hemavathy

Jagadeeshwaran

Jayanthi

Kamala

Lakshimipriya

226
 Libin

Madhu

Maria Neeta

Mohana

Mohana Sundari

Muralikrishnan

Nivetha

Parkavi

Poonkuzhale

Poornika

Premkumar

Ragavi

Raja

Renuka

Saravanan

Sathya

Shirley

Sorna

227
 Subash

Suriyaprakash

Vinothini

Executive Producer

Kannan Krishnamurty

NPTEL Co-ordinates

Prof. Andrew Thangaraj

Prof. Prathap Haridoss

IIT Madras Production

Funded by

Department of Higher Education

Ministry of Human Resource Development

Government of India

HYPERLINK "http://www.mptel.ac.in"www.mptel.ac.in

Copyrights Reserved

228
 NPTEL

NPTEL ONLINE COURSE

Introduction to Abstract

Group Theory

Module 05

Lecture 25 –“Problem 6”

PROF. KRISHNA HANUMANTHU

CHENNAI MATHEMATICAL INSTITUTE

(Refer Slide Time: 00:55)

Before we look at some examples, I want to make an important

remark. Cauchy’s theorem also holds when G is not abelian, so
you remove the word abelian in the theorem, so G is a finite group
and P is a prime number that divides the order of G then G
contains a element of order P, this is also true okay, this requires
more machinery to prove and we do this in later in the course.

So after I develop more machinery required to prove this, we will

prove this. In fact we will prove what are called Sylow theorems
which contain this and which in fact say much more.

So let me give you quickly an example to illustrate the fact that p

needs to be a prime number in the theorem. Consider the group, so
I am going to consider a few two by two matrices, so one zero

229
zero one, minus one zero zero minus one and one zero zero minus
one, minus one zero, zero one.

So G is a group of order four okay, so G is a subgroup of, here the

operation is multiplication, so G is a subgroup of GL2R, so
remember this all, note these all are invertible matrices. They have
all determinant one or minus one, so they are invertible matrices
it is subgroup of GL2. And fact that it is subgroup is clear, I will
give this name, so just to, for ease of working of this, let’s call this
as E and let us call this as a1, let us call this a2, let us call this a3.

(Refer Slide time: 02:52)

So G is identity element that I am calling E, right, what is a1

square? a1 square is e, that is this matrix multiplied with itself,
that is same as a2 squared that is same as a3 squared. What is
a1a2? That is if you multiply these two matrices you will see that,
it is easy to say that it is a3. a1a3 is a2, a2a3 is a1. So this is the
entire group of structure on this. The data, multiplication table is
completely given by this.

So G is order group in order four group, so G is an abelian group

of course, and in fact we have seen in a previous video that any
group of order of five or less is abelian , so this is certainly abelian
and you can see that from, I should written this, that is same as
a3a1 okay so that is easily verifiable from the given format, so
these matrices are given you multiply these matrices, if you do not

230
know the matrix theory, to see that how to multiply matrices and
so on, do not worry about it, just think about G as group like this,
with these operations, with this multiplication table, a1, a2, a3 all
have, G has one element of order one and three elements of order
two.

Okay, so as I said if you do not feel comfortable with matrix

representation, just do not worry about it, and think about G as a
group consisting of four elements E, a1, a2, a3 with this
multiplication table, they are, a1, a2, a3 have order two meaning,
their squares of all identity and multiplying any two of them will
give you the third, in any order, so a1a2 is same as a2a1 which is
a3. This is a valid group and it is a group of order four. G is called
Klein four groups. Okay, it is called the Klein four group.

Is it cyclic? Is it a power of, is it the group generate by some

specific element, it is not right, because it is certainly not group
generated by E, is it the group generated by a1, no a1 squared is
identity. So the group generated by a1 is just E a1, similarly group
is generated by a2 is e, a2, because a2 is also order two, same with
a3.

(Refer Slide time: 06:09)

So it is not cyclic, because it has no element of order four. Right,

in order for a group of order four cyclic, it must contain an
element of order four. In this case it does not, so this is another

231
group of order four, we have seen that there is a group of order
four namely the cyclic group of order four, this is not a cyclic
group of order four. And it is actually an exercise which I will do
maybe later, that up to isomorphism these are the only two groups
of order four, both are abelian, one cyclic and the other is not
cyclic. Now this is an important example to give you a various, to
illustrate various points.

(Refer Slide time: 07:04)

First of all, the converse of Lagrange’s theorem fails here, right,

the converse of Lagrange’s theorem, what is the Lagrange’s
theorem? If in an element, in a group, the order of an element
divides the order of the group. On the other hand, here four
divides order of G, which is of course, four, but G has no element
of order four.

Similarly this example shows that p in Cauchy’s theorem must be

prime right, because four divides the order of group G, G is a
finite abelian group, four divides the order of G, but G has known
element of order four.

The problem is 4 is not a prime. So only if a prime number divides

the order of a group, we can say there is element of that order, and
that of course checks out here, because order of the group here is

232
four, and the only prime that divides that is two, and certainly G
has a element of order two.

On the other hand if G is a cyclic group of order four, say G is

one, a, a2, a3. Then also G has an a element of order four, or
element of order two in, namely a2, Cauchy’s theorem applies to
this and says that if two divides four, two is a prime number it
must have an a element of two, so it does because it a square.

So in this case there is only one element of order two, because

order of a as well as order of a squared and order of a cube is two,
is four rather, order of a square is two, order of e, let call this
element as e, order of e is one. So the cyclic group of order four
has one element of order one, one element of order two, two
elements of order four. Whereas the Klein four group has one
element of order one, three elements of order two.

There is always in element of order two in a group of order four

by Cauchy’s theorem, in an abelian group of order four. But of
course every group of order four is abelian so there is no need to
separately say abelian but Cauchy’s theorem also say if you have
10 order 10 abelian group, it must have order two element and an
order five element.

(Refer Slide time: 10:13)

So here I am doing problems. Problem one, so let’s start with a

couple of very easy problems, just to get started, so let us say G be
a cyclic group and let H be a subgroup of G, then G mod H is
cyclic okay.

233
So first of all recall that G is cyclic implies G is abelian, in a
cyclic group is certainly abelian, implies H is normal in G, so G
mod H is a group, so in order to even say G mod H is a cyclic
group, first we need to know it is a group, but it is a group, so we
need to know observe that it is a cyclic group.

What is a cyclic group? Suppose G is generated by A, so G is

equal to A power n, n belongs to Z. So powers of A exhaust all of
G, so G is every element of G, is power of A, now what is G mod
H? G mod H is g times H, where g is in G, okay, these are the left
cosets of H in G.

(Refer Slide Time: 12:11)

So we claim that, what is your guess, we are trying to show that G

mod H is a cyclic group, if G is a cyclic group, G mod H is a
cyclic group. In the other words, we have to exhibit a generator for
G mod H, it is obvious to guess what? If G is generated by A, AH
generates G mod H, and this is easy, because you take gH, for g in
G, if g is in G right, g equals A power N, for some A. Right
because A generates group G, every element of G is a power of A,
so gH is equal to (A power N)H, but by the operations of cosets,
this is just AH power N, so G mod H, is equal to AH, power N, N
belongs to Z. Of course, we are not saying that they are all distinct
elements, some of them will collapse, if it is a finite group for
example, they will collapse. But that is irrelevant here, G mod H is
generated by AH, so this is the solution, so if G is cyclic then, then

234
G mod H is cyclic. And in fact the coset of the, element, generator
generates the quotient group.

(Refer Slide Time: 13:35)

Let us do one more problem which is very similar and also very
easy. If G is abelian, and H a subgroup of G, show that G mod H
is abelian okay. As I said this is also extremely easy, to show that a
group is abelian, you have to start with two elements of the group,
in this case the group we are interested in showing is abelian is G
mod H, so let us take two arbitrary elements, AH, and BH. These
are a two left cosets, what is the AH times BH. By definition of
the group operation on the quotient groups, this is ABH, but AB is
equal to BA, because your group is abelian and this is same as
AH, okay, so this is okay, this is also a trivial exercise, let us do
one more interesting exercise.

(Refer Slide Time: 14:39)

This is slightly more work, so more interesting. Suppose G, let ZG

be the centre of G, okay, if G mod Z is cyclic show that G is
abelian.

In order to first solve the problem, let’s first we call what is the
centre, remember ZG is all elements A in G, which commute
everything else in G. So A in G such that that Ag is equal to gA to
all g in G, so this is the centre. So now we are assuming that G
mod ZG is cyclic, we want to show Z is, G is abelian here, okay,
in order to show abelianness as in the previous problem we have

235
to take two arbitrary elements, and to show, we want to show,
show that AB equals BA, right let us keep this in mind, we want to
show AB = BA.

So now since G/Z(G) is cyclic, it has a generator, right it is

generated by some element of G/Z(G), right so this G/Z(G) is a
cyclic group it is generated by an element of that, say g Z(G), so
an element of this simply a left coset, so take a left coset and
suppose that it a generator, that means everything in G mod Z(G)
is a power of this, so in particular we have AZ(G), see AZG and
BZ(G) are elements of G/Z(G) so they are generated by gZ(G). So
it is some g(Z(G) power I and BZ(G) is some Z(G) power J, for
some integers I and J, okay that means because gZ(G) generates it
these two elements are power of it, so that means AZ(G)=(g power
I) Z(G) and BZ(G)= (g power J)Z(G), this is the meaning of coset
multiplication. But now let us pull this relation back to the group
G. So this means, so what is this mean? That means A times
something lets x, in the center is equal to g power I times
something, where x and x prime are in the center, right because A
times that, as cosets they are equal, that means A times an element
of this is equal to an element of this, so A times x will be equal to
this.

(Refer Slide Time: 19:07)

So I can rewrite this as A= g I power X for some X in Z(G), okay

so this X and X prime I am interchanging so it is actually this X

236
here is this x prime times x inverse because X and X prime are in
the center, X prime times X inverse is also in the center, so I am
renaming everything to call it X. Similarly we have B = g power J
times Y for some Y in the center, so far fine, right, A = g power I
times X for some X in the center, similarly B = g power J times Y
for some Y in the center. Now let us go ahead and calculate AB,
now that we have written everything in terms of elements of
capital G and not as cosets, let us write AB.

So this is g power I X g power J Y, but now what is X and Y, X

and Y are in the center, so they commute with everything, so this
is same g power I, g power j, X, Y right, this is g power I +J, XY,
on the other hand, what is BA? This is g power J, Y g power I, X
again using the fact that X and Y are in the center, AB=BA, so I
am using the fact that g power I and g power J is g power I +J and
X and Y are in the center. So this shows that because A and B were
arbitrary elements of the group and they commute, G is abelian
okay.

(Refer Slide Time: 22:00)

So if G mod is the center cyclic, G is abelian.

So the next problem and this is the problem that I used in the
proof of Cauchy’s theorem. It says the following. Let G be a cyclic
group. Suppose that, which is finite so this is finite cyclic group.
Suppose that M divides, M is a positive integer, suppose M is a
positive integer that divides the order of G, okay, so then show
that G contains an element of order M. So that is the problem, so
solution, so this is problem 4 I think okay.

237
Solution is the following. So suppose that G is generated by A
right, so G is finite cyclic group okay, so G must be of the form E,
A, A2 , A power (N – 1), where N is the order of A as well as order
of the group, right so G is like this so G has N elements and order
of A is N. And if M divides consider the element A power N /M.

(Refer Slide Time: 24: 01)

So consider this element, A power N/M. See that because M

divides N and M is a positive integer remember M is a positive
integer, N by M is also an integer so A power N by M is a valid
element of the group, it is A multiplied by itself N by M times, so
A power N by M power M is of course A power N which is
identity, remember as I have repeatedly said in the past, you can
apply the usual exponential rules to group operations.

So A power N by M, if you call this element B, B power M is E,

can we say order of B is M? If so, we are done, right, because we
are trying to show that G contains an element of order M, if order
B is M we are done, but actually we cannot right away say this, in
order to say order of B is M, you have to say, no not yet we can’t
say this yet, what is order of B? Order of B is the least positive
integer D such that B power D is E, right, certainly M is a
candidate for it B power M is E, but we have to rule out the
possibility that something less than M has this property.

(Refer Slide Time: 27:04)

238
So if B power D is E, so let us say, for suppose for some D < M,
we have B power D is equal to E. Then let us see what are the
consequences of this, B power D is E, this means A power N/M,
so remember A power N/M is DB, power D is E, right this means
A power (ND /M) is E but if D is less than M , ND/M is less than
N correct, because D by M is less than 1, but this contradicts, we
have a power of A equal to E but that power is less than N, we
have a contradiction, we have contradicted the fact that order of A
is N, right.

(Refer Slide Time: 29:40)

Because order of A is N means N is the smallest positive integer

such that A power N is identity, but here we are getting A power
(ND/M) is E and ND/M <N. So this a contradiction and only
assumption that we could have possibly made incorrectly was that
D is less than M, so we must have, so A power M identity and A
power D is identity implies that D is at least N, so order of B
power M I, sorry I should not have said A power M, B power M is
identity where B is A power N/M, B power M is identity and no
smaller power of B is identity. So order of B is exactly equal to M
and this completes the solutions of the problem because we want
to show G contains element of order M and we have shown that
because B is that element okay.

I will only remark one more time that we have seen the example
of Klein 4 group, where you have a group of order 4 yet the group
does not contain an element of order 4, of course this is also

239
illustrated by the symmetric group on three letters S3, S3 has 6
elements, but it not cyclic.

So it has no element of order 6, so the statement that G contains an

element of order M if M divides the order of G is simply false if G
is not cyclic. So let me remark, it is crucial that we assume G is
cyclic in this problem, because example take G to the Klein four
group or G is S3, this has order of 4 and this has order 6, but they
are not cyclic, not cyclic, and they have no element of order 4 or
order 6 respectively, okay. I will stop the video here, hopefully
these problems gave you some idea of how to apply various
results that we have learned and you should try to do more
exercises that I have been giving in the videos using these ideas.
Thank you.

Online Editing and Post Production

Karthik

Ravichandran

Mohanarangan

Sribalaji

Komathi

Vignesh

Mahesh Kumar

Web Studio Team

Anitha

240
 Bharathi

Clifford

Deepthi

Dhivya

Divya

Gayathri

Gokulsekar

Halid

Heamvathy

Jagadeeshwaran

Jayanthi

Kamala

Lakshmipriya

Libin

Madhu

Maria Neeta

Mohana

Mohana Sundari

241
 Muralikrishnan

Nivetha

Parkavi

Poonkuzhale

Poornika

Premkumar

Ragavi

Raja

Renuka

Saravanan

Sathya

Shirley

Subash

Suriyaprakash

Vinothini

Executive producer

Kannan Krishnamurty

NPTEL Co-ordinators

242
 Prof.Andrew Thangaraj

Prof. Parthap Haridoss

IIT Madras Production

Funded by

Department of Higher Education

Ministry of Human Resource Development

Government of India

HYPERLINK "http://www.nptel.ac.in" www.nptel.ac.in

Copyright Reserved

243
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 05
Lecture 26- “Symmetric group 1”
Prof. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

In the course so far we have studied many properties of groups,

most recently we have looked at isomorphism theorems, quotient
groups and as an application of these we have studied the
Cauchey’s theorem which says that if a finite group has order (N)
and you have a prime number that divides (N) then there is an
element of that prime order.

The next important topic in the course is symmetric groups; I am

going to spend this and probably the next video studying
symmetric groups. We already looked at symmetric groups as an
example and we looked at one of these groups in detail. So what
are symmetric groups. Lets recall, so let N be a positive integer
and we denote the symmetric group on N letters by (SN) so SN
denotes the symmetric groups on N letters, what does this mean?
So this means the following: so we usually give the letters we call
by 1 to N, so (S N) is the group of all bijections of an (N) element
set. So you take a set with (N) elements and you consider all
bijections of that set, usually we denote the set which has (N)
elements, when I say an (N) element set I mean a set having (N)
elements. It has (N) elements right, it is very natural to give names
to these (N) elements, so let’s call them 1, 2, 3 up to (N).

244
So, these are the (N) elements, do not think of them really
numbers. It is sometimes useful to maybe think of them as
numbers, but the point is there are (N) elements in that set which
we are denoting for convenience as 1,2,3 up to (N). So we already
are familiar with S3, this is one of the groups that we have been
looking at in various situations in we saw. How many elements it
has and so on, so for example what is the order of S 3? 6 and we
have various ways of writing this, so, I introduced in the very
beginning one notation where I called the elements so maybe I
called them f1 through f6 and this we thought is not very useful, it
is not convenient because we do not know what f 1, f2 are and we
have to go back and see what they are and because of that I
introduced a notation and we will discuss this in more detail today.

This notation that I introduced, so, I am going to spend this video

explaining how to understand (S N) and what this notation is and
what are its properties and we will consider some basic properties
of the symmetric group.
(Refer Slide Time 04:26)

So symmetric groups, remember, just not one group, it is a class of

groups, for every positive integer we have one group for example
let’s quickly settle what S1 and S2 are.

What would be S1? S1 is by definition bijections of a one element

set, so and what are they, there is only one bijection right, there is
only one bijection from a single element set, remember we call the
element 1, there is no surprise here (1 has to go to 1), so it is the

245
identity bijection. Remember E denotes the identity element of the
symmetric group and as a function this stands for the identity map
which is definitely a bijection. And I should remark here, of
course, that when I say group of all bijections I mean under
composition. I way back in one of the first videos we saw that
bijections under compositions form a group.

So, the binary operation in this group is composition. Every

bijection has an inverse, identity bijection serves as the identity
element and composition of two bijections is a bijection, so this is
a group under composition. So, S1 is the trivial group just
containing one element.

What about S2? So we have to look at bijections from (1,2) to

(1,2). So there is the identity bijection so (1 goes to 1, and 2 goes
to 2), this is there always and we have another bijection. It sends
(1 to 2) in order to be different from this 1 must go to some
different from 1, so 1 goes to 2 and 2 goes to 1. So let’s call this
(12) following our notation I am going to recall this again for you
later in the video (12) remember is the bijection that sends (1 to 2,
2to 1) and there are no other elements here, no other letters here.

So, it is just E and (1,2), and of course, S3 I have written here, is

these six elements so, the first thing we want to understand is and
of course S4 will have more elements. It will be bijections of
(1,2,3,4) and S5 will be bijections of (1,2,3,4,5) and so on, so first
thing is, first question let us settle this first, what is the order of (S
N)?

246
So here we are looking at bijections, so it is a simple set-theoretic
count. So (SN) is all bijections from (1, 2 up to N) to itself, so
these are bijections, lets think for a minute on how to construct a
bijection. Bijection is a function which is 1-1 and onto. So 1 has to
go to something, 2 has to go to something, similarly 3 has to go to
something, finally (N) has to go to something. First let’s see how
many options there are for 1, there are (N) options, because 1 can
go to 1, 2, and 3 or (N), it can go to any of the (N) letters, so there
are (N) options. Now how many option are there for 2, now for 2,
remember in order to be a bijection 1and 2 be distinct letters must
map to distinct letters, so, we have made a choice for 1 in the
beginning, 2 cannot go to that, if 1 goes to 3, for example 2 cannot
go to 3, but 2 can go to anything else, so you remove wherever 1
has been sent, and 2 can go to any of the remaining ones, that
means there (N-1) options, on where 2 can go, 3 now can go to
any of the letters which are different from images (1 and 2) so 3 as
(N-2) options, 4 as (N-3) options, (N) finally after you do all this,
there is only one option for this, and this is the principle at work if
you remember how we worked out S3, S1, S2.

So, you can send 1, so (1,2,3 up to N) is an arbitrary order, we fix

them in some way, which order is irrelevant, it is convenient to
think of 1 as the first element, 2 as the second element, and so on.
So 1 goes to any of the (N) options, 2 goes to any of the remaining
(N-1) options, 3 goes to any of the remaining (N-2) options, (N)
goes to after having decided the images of 1,2,3 up to (N-1), we
have exhausted (N-1) elements of the set 1,2 up to (N), so there is
only one option left for (N). Now how many there total? So this is
the principal in counting if how many total number of bijection are
there.

247
So the total number of bijections, because 1 can go to (N) of them,
it is (N) times 2 can go to (N-1) of them, 3 can go to (N-2) of
them, (N-1) can go to 2 of them, so this is the product so, (N-11)
times (N times N-1 times N-2 all the way up to 2 times 1) and
usually we call this by (N factorial). Read this as (N factorial), so
the total number of bijection is (N factorial) because where you
send 1 is independent of where you send 2, apart from making
sure that 2 has to go to something that is not already the image of
1, once you keep that in mind, 2 can go to anything, so you can fix
one of the options for 1 and you can construct a function, so now
change the option for 1 and construct same number of functions
and so on, so the total number of bijections is (N factorial). Hence
the order of (S N) is (N factorial).

So this is the first important property of (S N). (S N) is a group of

order (N factorial), and clearly check that this checks out in cases
1, 2, and 3, S1 should be one factorial, which is 1, and that is
correct, S2 should be 2 factorial which is 2, which is also correct,
S3 order must be 3 factorial which is 6, which is correct, so (S N)
is a group of order (N factorial). So the size is increasing, it gets
very big very fast, for example S4 as order 24, S5 as order 120, S6
as order 720 and so on. So these orders increase very fast because
each time you multiplying by (N) so this is the size of the group.

Now how do we express elements of (S N), so this is the next

question, how do we express elements of (S N)?

248
So there are several ways of expressing elements of (S N), we
want to choose one which is most convenient in the sense that it is
efficient we do not need to write lot of text and it is easy to work
with in terms of multiplication of , in terms of figuring out
whether an element is in a subgroup, things liKe that. So how do
we express elements? So I want to introduce to you and this is
actually introduced to you when we discussed S3 earlier, but I will
do this again in more systematically. We use cycle notation. So
hopefully this is already familiar to you from the time we
discussed S3, so and in any case I will discuss all the details again.
So what is the cycle notation?
(Refer Slide Time 14:02)

So it is best explained in terms of, with an example. So let’s look

at, for example, let’s start with this. So before I do the example,
just a notation so that I can keep using this, elements of (S N), are
called, are usually, permutations, instead of using the word
bijection, lets use the word permutation. Permutation is just
another word really for bijection, it is about, if you take (N)
elements, how do you permute them how do you map those (N)
elements to themselves again, so that is really a bijection, but this
is more mathematical term so we use the word permutation.

So let’s take a permutation and I am going to denote this by

sigma, and I am going to use these Greek letters, so sigma is an
element of S9, so it is a permutation on (9) letters, so what is this?
Lets use one particular way of writing this, so let’s use this
notation, so let’s use this as a matrix, represent this as a matrix.
(Refer Slide Time: 19:52)

249
Where the first row as all the nine elements, of this set (1,2,3 up to
9), so sigma is a permutation of it, so sigma sends (1 to 3), lets do
this (2 to 4). So the top row represents the letters the bottom row
represents where they are mapped under sigma. So sigma maps (1
to 3), (2 to 4), (3 to 2), (4 to 1), (5 to 6), (6 to 5), (7 to 8), (8 to 7),
(9 to 9), so this sends for example (3 to 2), (6 to 5) and so on, I
hope the notation is clear. 1 goes to 3, 2 goes to 4, and so on, so
clearly this is nice, clearly it tells you what element sigma does, in
the sense that what is the function, remember elements of (S N)
which I am calling permutation are really functions.

They are functions from (1,2,3 to N) to (1,2,3 up to N), so in that

way sigma is clear, what as a function it does is clear, (1 goes to 3,
2 goes to 4) but certainly you can see that, there is some
inefficiency in expressing in sigma, for example the top row is
unnecessary. Because we know that it is always 1, 2, 3 up to N. So
but if you remove that it is somewhat difficult to keep track of
what the function is, so now rewrite this in the following way.

Rewrite sigma in the following way, so what is the way that we

want to think of sigma is the following. So let’s start with 1, so I
can start with anything, let’s start with 1, see where it goes under
sigma. So I am starting with 1, see where it goes, it goes to 3, and
see and repeat till, so let me write that first, so see where it goes
under sigma, put it next to, and repeat until we come to 1 again, so
that is what we do.

So let’s illustrate this, we will start with 1, see where it goes it

goes to 3, repeat with 3, so see 3 goes to 2, it is right there, the
function sends 3 to 2. We have not come to 1 again, so we will
continue. 2 goes to 4 under this, again we have not reached 1, so
continue. Where does 4 go? 4 goes to 1, so 1 we have come again.

250
So we close the bracket. So this is the, we call this, the cycle
determined by 1. We call this the cycle determined by 1. Again let
me repeat the procedure. We start with 1, see where it goes, put it
next to it, so 3, see where 3 goes, it goes to 2, put 2 next to 3,
where does 2 go, it goes to 4, so put 4 next to 2, and 4 goes to 1,
so you close the bracket. You should think of this as a cycle. So
because you start with 1, you go clockwise, go to 3, then you go to
2, then you go to 4, then you come back to 1, okay.

Of course, this whole thing can start with 2 also. If we started with
2, then what happens? See where 2 goes, 2 goes to 4, 4 goes to 1,
1 goes to 3, 3 goes to 2, so we close the bracket. It is the same
cycle. On a circle it doesn’t matter where you start, right.

So instead of 1 I will start here, so that is (2, 4, 1, 3), if I started

with 3 it will be (3, 2, 4, 1), if I started with 4 it will (4, 1, 3, 2), so
this is one cycle, sigma is not just this.
(Refer Slide Time: 20:16)

So now in the next step, this is the step 1, step 2, find an index, if,
that is not contained in the cycle determined by 1, find an index,
so let us call 1, 2, up to N indices, okay so the set is 1, 2, 3,…N,
the elements of it are called indices, elements of the symmetric
group are called permutations, so permutations permute indices.

So find an index is not contained in the cycle determined by 1, so

1, 3, 2, 4, are contained in the cycle, let’s pick 5. You can also
chose 6 and also 7, it does not matter, so I am just choosing 5, now
find the cycle determined by 5. So in general, we have learned in
the step 1, how to find the cycle determined by an index, in this
case let us do it for 5.

251
So you put a bracket and start with 5, under the function where
does 5 go, 5 goes to 6, now where does 6 go? 6 goes to 5, so I
close the bracket, so I get (5 6), so it is a cycle, so 5 goes to 6 and
6 goes to 5.

So next, you see now what happens, find an index not contained in
cycles constructed so far, remember so far we constructed two
cycles, (1, 2, 3, 4) and (5, 6), so find an index that is not contained
so far, so let us choose 7, and determine the cycle, find the cycle
determined by 7. It would be 7 and in the function, what is the
image of seven, 7 goes to 8, so 7 goes to 8 and 8 goes to 7, so you
take (7, 8) and you close the bracket, this is the third cycle.
(Refer Slide Time: 23:09)

So this is the third cycle, and last cycle will be, find a cycle, so
again is the same idea now, find an index not covered so far,
which is only 9, 9 is the only index that is not covered so far, and
the cycle determined would be in the function, 9 goes to 9, so you
just use the index, so you have only 9 in this cycle so, you have
the single element so, we say that, we can write sigma equals
(1324) was the first cycle, (56) was the next cycle, (78) is the third
cycle and, (9) is the last cycle. We usually do not write a cycle, if
it has only one index okay, because if it has one index, it is clear
that that index will goes to itself, so we usually omit writing 9, so
we write as sigma as (1324), (56), (78), so the same function, the
same function we have has this cycle decomposition.

(Refer Slide Time: 24:41)

252
This is called the cycle decomposition of sigma, so what are we
doing? We are decomposing sigma as cycles, so a cycle, first of
all, is an element, which starts somewhere and ends with that, so 1
goes to 3 in the cycle 3 goes to 2, 2 goes to 4, so cycles are
permutations of the form, so now let me introduce a general
notation here okay, so this is a cycle, it has K in this cycle, so this
is a K-cycle, so we call it as K cycle, what is a K-cycle? It is a
cycle with K indices, what does this K-cycle do? What does the
K-cycle (i1, i2, i3,…, iK) do?
(Refer Slide Time: 26:16)

It sends i1 to i2, i2 to i3, i3 to i4 right, you understand, i1 the first

thing goes to i2, which is right next to i2, i2 goes to i3, which is
right next to i3, right next i3 is i4 so, i3 goes to i4 and so on, i(K-
1_ goes to iK, because i(k-1) comes just iK, so, i(K-1) goes to iK,
and finally where does iK go? Because it is a cycle, you cycle
back there is nothing next to iK means, you send iK to i1, so as a
cycle you have i1, then i2 always clockwise i4, i5, …, i(K-2), i(K-
1), iK, right, so i1 goes to i2, i2 goes to i3, i5 goes to i6, i6 goes to
i7, i(K-1) goes to iK, i(K-2) goes to i(K-1), i(K-1) goes to iK, iK
goes to i1 so this clearly says that, this same cycle can be written
as (i2,i3,i(K-1),iK,i1) right, so start with i2 and end with i1, and
you can write them with any of these options, you can start with
i3, you can choose any of the k indices as the starting point.

(Refer Slide Time: 28:00)

But now the description of the cycle is not done, I have only told
you what happens to these indices, further I need to tell you what
happens to indices that are not in this set, if j is an index not
contained in { i1, i2, …..iK}, see i1, i2,…., iK are just some

253
indices, which are, remember this is a subset of, so if j is an index
not contained in this i1, i2,…,iK, what does the cycle { i1, i2,
…..iK} do to j, it sends j to j, it does not do any thing to j, now I
have completely described the cycle.
(Refer Slide Time: 28:59)

As an example consider the cycle, let us say (1, 2, 3) in S6 what

does it do, as a function (1, 2, 3) is given by, S6 is the symmetric
group of 6 letters.
(Refer Slide Time 29:36)

So I need to tell you what is the image of those indices under (1, 2,
3). 1 goes to 2, 2 goes to 3, 3 goes to 1, so all the indices contained
in the cycle behave this way, and all the indices that are not
contained in the cycle are fixed okay, so we say (1, 2, 3) fixes 4, 5
and 6.

So more generally the cycle (i1, i2, …,iK) fixes any index that is
not contained in i1, i2, …., iK okay. So now going back to the
example we worked out here, this particular permutation sigma
can be written as a product of cycles (1, 3, 2, 4), (5, 6) and (7, 8).
(Refer Slide Time: 30:36)

So in our example above, sigma which was in S9, can be written

as a product, okay, of disjoint cycles, okay, let me explain this in a
minute, in the following way: (1, 3, 2, 4) (5, 6) (7, 8). These are
disjoint cycles because they have no common indices, the indices
that (1, 3, 2, 4) contains are 1, 3, 2, 4, (5, 6) contains 5, 6, (7, 8)
contains 7 and 8, so there is no common index, so they are disjoint

254
and they are cycles. So sigma can be written as a product of
disjoint cycles.
(Refer Slide Time: 32:19)

Okay now if you think about the procedure of how we did this for
this particular sigma, hopefully it will be clear to you that we can
to this for any sigma, the same procedure that we used for sigma
in S9, can be applied to any element in any symmetric group, so
we can conclude with a proposition.

(Refer Slide Time: 32:51)

Every element sigma in a symmetric group has, a cycle, has a

decomposition as a product of disjoint, right, remember that this
clear, because you start with sigma, ask where 1 goes, I am just
recapping the procedure that we followed, start with 1, determine
the cycle determined by 1, start with 1, see where it goes, you put
it next to 1, see where that goes, put it next to that and so on. At
some point obviously you will reach 1, because the SN is the
symmetric group on N letters, only N indices are there, so you
cannot keep forever going without reaching one. So at some point
we must reach 1, then you close the bracket, and that is the cycle
determined by one, may be you exhausted all the indices, then you
have a 9-cycle, all the 9 indices are in the cycle, or maybe you
have not exhausted all N indices, N is in general now. So it is
possible that you have all N indices in 1 cycle then you are done,
element itself is an N-cycle.

Otherwise there is an index that is not contained not in the cycle

by determined by 1, then you start with that, and see the cycle
determined by that, and you keep doing this, because it is all on N

255
letters, you must have at some point exhausted all the indices, so
you are done. So every element of SN has a decomposition as a
product of disjoint cycles.
Online Editing and Post Production

Karthick

Ravichandaran

Mohanarangan

Sribalaji

Komathi

Vignesh

Mahesh kumar

Web Studio Team

Anitha

Bharathi

Catherine

Clifford

Deepthi

Dhivya

Divya

256
 Gayathri

Gokulsekar

Halid

Hemavathy

Jagadeeshwaran

Jayanthi

Kamala

Lakshimipriya

Libin

Madhu

Maria Neeta

Mohana

Mohana Sundari

MuraliKrishnan

Nivetha

Parkavi

Poonkuzhale

PoorniKa

257
 Premkumar

Ragavi

Raja

Renuka

Saravanan

Sathya

Shirley

Sorna

Subash

Suriyaprakash

Vinothini

Executive Producer

Kannan Krishnamurty

NPTEL Co-ordinates

Prof. Andrew Thangaraj

Prof. Prathap Haridoss

IIT Madras Production

Funded by

258
Department of Higher Education

Ministry of Human Resource Development

Government of India

HYPERLINK "http://www.mptel.ac.in"www.mptel.ac.in

Copyrights Reserved

259
NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract Group Theory
Module 05
Lecture 27 –“Symmetric groups II”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

I will do, hopefully the proof is clear, and I will do just couple of
more examples.
(Refer Slide Time: 00:21)

You take S5, so let us use this inefficient, but clear description of
an element, let us say sigma is, 1 goes to 2, 2 goes to 3, 3 goes to
4, 4 goes to 5, 5 goes to 1. So what is the cycle decomposition? So
this is called cycle decomposition.

So as I said, start with 1, 1 goes to 2, 2 goes to 3, 3 goes to 4, 4

goes to 5, 5 goes to 1, so sigma is a 5-cycle, so that is all, because
you have exhausted all the indices, there is no further cycle, in
general you will have a product of several cycles. In this example
sigma was an element of S9, it is a product of 3 disjoint cycles.
See when I say product I mean I have told you how to apply
product of cycles earlier.

In this case, we start with 7 and see where it goes? It goes to 8.

Because it is disjoint thing 8 will not appear again, so 7 goes to 8,

260
8 goes to 7, 5 goes to 6, 6 goes to 5, 4 goes to 1, 1 goes to 3, 3
goes to 2, and 9 is not represented here, because 9 goes to 9 itself.
In this example, sigma is a 5-cycle, one more example, so let us
take.
(Refer Slide Time: 1:57)

Let us take S5 again, let us take tau to be 1 goes to 1, 2 goes to 2,

3 goes to 4, 4 goes to 3, and 5 goes to 5. So what is the cycle
decomposition of tau? So this is called tau. Tau is this, so you start
with1, 1 goes to 1, so you have, that is just a 1-cycle. 2 goes to 2
again it is a 1-cycle, 3, now you start with 3, which is an index not
covered yet. 3 goes to 4 and 4 goes to 3. So you close the bracket
and finally 5 goes to 5. So this is stopped.

But remember our convention, we don’t write, because 1-cycle is

unnecessary to write, there is so information you gain by writing
1, 2 and 5. So, see tau is just a 2-cycle. In this case tau is a 2-cycle.
So now the general proposition is every element can be written
like this. Every element has decomposition as a product of disjoint
cycles. Another property of this, of disjoint cycles, which is also
clear, another property of disjoint cycles is that if, sigma and tau
are disjoint cycles, then sigma tau equals tau sigma, that is in
words, sigma and tau commute with each other.
(Refer Slide Time: 04:00)

See, remember that in general symmetric group is not abelian, S3,

we have seen in example detailed description of S3 earlier, and we
know that S3 is not commutative , so in general two elements of
S3 do not commute with each other. For example, so recall, in fact

261
SN is not abelian, if n is greater than or equal to 3. So this is an
exercise in fact. S3 and S4, S5 they are not abelian they cannot be
abelian, S1 and S2 are obviously abelian, because S1 is just a
group with one element, S2 is a group with 2 elements, any group
of order 5 or less we saw is abelian. So those are abelian and in
fact without that exercise it is clear that a group of 1 and orders in
that exercise also it was very easy to conclude that group of order
1 or 2 is abelian.
But SN is not abelian, for example if you take (1,2) and (1,2,3).
What is this? If this is sigma and this is tau, so remember how do
we multiply, we start on the right side cycle, 1 goes to 2 and 2
goes to1. So 1 goes to 1 in this product, 2 goes to 3, and 3 goes to
3 under that, so 2 goes to 3, 3 goes to 1 and 1 goes to 2. So we
close the bracket and again we don’t write 1 cycle, so this (2,3).
On the other hand, what is tau sigma? (1 2) 3, (1 2). So in this case
1 goes to 2 and 2 goes to 3, so 1 goes to 3. And 3 goes to 3 in this
but 3 goes to 1 in this. So you stop there, 2 goes to 1, 1 goes to a2.
So again we don’t write the 1 cycle 2 consisting of 2 so we have
(2, 3) and (1,3) and these are different.

These are different, so ((1 2)) times (12 3) is not same as ((1 2) 3)
times ((1 2)). So in general SN is not abelian, however what did I
say here?

If they are disjoint cycles, they commute with each other. In this
case they are not disjoint cycles, ((1 2)) and (123) are not disjoint.
Because the index 1 appears in both the index 2 appears in both,
so they are not disjoint cycles, however if they have the same, if
they do not have any common indices, we must have that they
commute with each other.

262
So now let me prove that, if sigma and tau are disjoint cycles, so
however so in general they don’t commute with each other, but if
they are disjoint cycles, then sigma tau is tau sigma. So why is
this?

If you think about it, it is clear, because when you try to write the
product, what do we do? We first look at indices that are in sigma,
and you see where it goes, but we first see indices in tau and see
where it goes. Wherever it goes it won’t appear in sigma again,
unlike in this example in this case 1 goes to 2, but 2 goes to 1,
because 2 appears here, but if there is no index that is common to
both sigma and tau, whatever happens in sigma stays within
sigma.

So when we consider, the product sigma tau we look at indices in

tau first. Because they are not present in sigma, because they are
disjoint cycles, this is very important, because they are not present
in sigma, it does not matter whether we consider their images, we
consider them first or after sigma. This is just something that
hopefully is very clear trying to write it, it may be necessary, but
let us just illustrate this.
(Refer Slide Time: 09:13)

For example if you have (1 2) and (3 4) , this is sigma and this is

tau. If you have (1 2) and (3 4), what I am trying to indicate here
is, so look at the indices in tau, 3 is the first index, 3 goes to 4 but
4 is not present in (1 2). So 3 goes to 4 under the product also,
similarly 4 goes to 3 under tau and 3 is not present in this. So 3
goes 4 , 4 goes to 3. Similarly 1 goes to 2 under sigma and it is not
affected by what happens in tau so 1 goes to 2 and 2 goes to 1, so
we might as well write this as, which is tau sigma.

263
If the indices are disjoint they are unrelated they don’t interfere
with each other, so we have no problem and we can just multiply
them in any order, so the cycle decomposition, this means in the
cycle decomposition of a sigma, I have already told you that every
element has a cycle decomposition. Now I am telling you that in
the cycle decomposition of an element sigma the cycles can
appear in any order.

In the example above, in SN remember SN, we discovered that it

is (1, 3, 2, 4), times (5 6) times (7 8). But now I am saying that
there is no problem if we write it like this. (5 6) (1, 3, 2, 4) (7 8)
or (7 8) (5 6)(1, 3, 2, 4), (7 8) (1, 3, 2, 4) (5 6), they are all same.
That is because the indices are different, are disjoint, so whatever
happens within (1, 3, 2, 4) , does not interfere with what happens
with (5 6). And with what happens with (7 8). So I can blue in the
order.
So the two important properties of cycle decomposition hat we
have learned so far is, every element of SN has a cycle
decomposition is, and the order in which cycle decomposition we
write element is irrelevant. So order in which we write is not
important, we can write it in any order. So some other properties
of cycle decompositions.
(Refer Slide Time: 11:35)

So consider a k-cycle, use SN, let sigma be a k-cycle in SN, so

we can write, sigma as, remember sigma can be written as (i1, i2,
…, ik). So I must stress here that not every element is a k-cycle by
itself, for example this element of S9, is not a cycle, it is a product
of 3 cycles. So a cycle is just one cycle, so its cycle decomposition
is itself, those are very special elements in SN. Sigma is a k-cycle

264
means it is a single cycle, but in general elements of SN can be
products of more than 1 cycles, in this example sigma is a product
of 3 disjoint cycles.

Now I am considering a single k-cycle, I want prove the

proposition that the order of sigma, in other words, order of a k-
cycle is k. So recall what is an order of an element? In general
order of sigma would be by definition the smallest positive integer
d, such that sigma power d is identity element.

So we want to show that sigma power k is identity, and sigma

power any smaller thing is not identity. So in order to do, that let
us find out what are the powers of sigma. So sigma is 9i1, i2 up to
ik). And recall what the meaning of this is? It sends i1 to i2, i2 to
i3, so on finally i(k-1) to ik. And what happens to an index that is
not inside i1 to ik, nothing happens, it is fixed by sigma. If a j is
different to i1, i2, to ik it sends j to j, so we don’t need to worry
about it, j goes to j for any j that is different from i1 to ik.

So now what happens to sigma square, what is sigma square?

Remember sigma square, because the operation in the symmetric
group is the composition, is sigma composed with sigma. Where
does i1 go under this, so i1 goes to i2 under sigma under another
sigma it goes to i3. So i1 goes to i3 under sigma square, so I am
going to keep track of the cycle of i1 under sigma , so i1 goes to i3
so I write it like this so (i1, i3. i1 next to is i3.

What happens to i3, let us figure out what happens to i3? Under
sigma i3 goes to i4, and what happens to i4 under sigma it goes to
i5, so i3 goes to i5. Similarly i5 goes to i6 then i7, i 5 goes to i6
under sigma and under another application of sigma i6 goes to i7.

265
So i5 goes to i7. So it keeps on going like this, which element
goes to i1? So of course here I should write ik goes to i1, under
sigma i(k-1) goes to ik under sigma, ik goes to i1 under sigma, so
you have i(k-1) here. i(k -1) goes to i1. What happens to ik it must
appear in the previous place somewhere in the middle.
(Refer Slide Time: 16:07)

So what we have concluded is sigma square is another, k-cycle.

Remember sigma square is an element of the symmetric group. So
it has a cycle decomposition, in fact it is a k-cycle itself. Because
where is ik appearing here? See i(k-2) goes to i(k-1) under sigma
under again sigma it will go to ik. So ik will appear here. So every
element i2 will also appear, i2 will go, where does i2 go? ik goes
to i1 and i1 goes to i2, so ik goes to i2. So they all appear here.

For example if I take sigma to the (1 2) (5 7), it is inside S7. What

is sigma square? Sigma square is it starts with 1 and I just skip one
step, so 1 to 2 and then 5, so it is (1 5). 5 goes to 7 under sigma so
7 goes to 1 so this (1 2 5 7), so 1 goes to 2 2 goes to 5, and 5 goes
to 7, and 7 goes to 1, so 5 actually goes to 1.
2 goes to 5 and 5 goes to 7. So 2 goes to 7, so 7 goes to 1 and 1
goes to 2, so actually you can see in this example what I wrote
here is just wrong. So sigma squared is not in general a k-cycle, as
this example shows.

I should have not written that, so sigma squared is not in general a

k-cycle, because may be not all indices are covered there, as this
example shows, so sigma is a 4-cycle, but sigma squared is a
product of two 2-cycles. Sigma is (12 5 7), sigma square it is (1 5)

266
(2 7). But doesn’t matter, so sigma squared sends in any case,
sigma squared sends i1 to i3.
What happens to sigma cubed? So I am going to keep track of
what happens to i1 under each successive power of sigma, sigma
sends i1 to i2, this is the first step, sigma squared sends i1 to i3,
sigma cubed sends i1 to i4, remember the fact that we are saying
sigma is a k-cycle, implicitly means that if it is a k-cycle, it means
that i1, i2 and ik are distinct indices.
(Refer Slide Time: 19:39)

Otherwise there will not be k of them so I won’t call it a k-cycle,

and more than that, in a cycle we are not allowed to repeat
elements , this is not a cycle ,because if 2 goes to 2, but 2 also
goes to 1 under this, in a cycle we cannot have a reputation of an
index, so if it is a k-cycle they are all distinct indices.

So now let us come back to this, sigma sends i1 to i2, sigma

squared sends i1 to i3, sigma cubed sends i1 to i4. And you can
see that sigma k-1 sends i1 to 3 goes to 4, 2 goes to 3, so this goes
to ik. And this is not surprising, so I have to repeat k-1 times we
apply this, i1 will go to the last index.

So since i2, i3 up to ik are all distinct from i1, remember that that
is the point of being a k-cycle, this is not i1, this is not i1, this is
not i1, this is not i1, they are all not distinct. Because (i1, i2, i3 ..
ik) is a k-cycle, and sigma can’t be identity, because sigma sends
i1 to i2 which is different from i1, so sigma cannot be identity. I
don’t care about what it does to other elements, it is a i1 to i2, so
identity element is supposed to send i1 to i1, but sigma sends i1 to
i2, so sigma cannot be e.

267
Sigma square sends i1 to i3 so it cannot also be e. Because i3 is
different from i1. So identity element sends i1 to i1, similarly
sigma cubed can’t be e and finally sigma k-1 cannot be e. So
sigma is not e , sigma squared is not e, sigma cubed is not e, sigma
k-1 is not e. So order of k, order of sigma, has to be at least k. So
remember order of sigma is the least positive integer d, such that
sigma power d is identity.

So here sigma is not identity sigma squared is not identity, sigma

k-1 is not identity. So the order has to be al least k. Is it k? Let us
find out what sigma k does. Under sigma k, i1 goes to i1, because
under sigma k-1, i1 goes to ik, if you apply sigma again, it goes to
i1. So far so good, sigma k has the chance to become be the
identity element, because i1 goes to i1.
(Refer Slide Time: 23:12)

What about sigma k i2, so i2 remember, under sigma goes i3,

under again sigma it goes to i4 and if you keep applying this, at
some point you will get ik and then i1, then i2. So if you do k
times, so sigma k of i2 also i2. Because it is the same, it is a cycle,
and as I said cycle can start with any of the indices in it, so the I
said here with i1 but (i1, i2, ik) can be written as (i2, i3, ik-1, ik,
i1). And now i2 is the first one, in the previous calculation they
also said the first one after k times we will reach the first one
again.

So i2 after applying k times, we get to i2 again, so similarly, sigma

k power i, any of them , so sigma is i3, sigma k power i4 is i4, and
so on and finally sigma k power ik is ik. So we have sigma k is the
identity element, right, because sigma k times i1 to i1, i2 to i2, i3

268
to i3, ik to ik but does it another indices which are not inside i1
through ik to themselves? Of course it does, because sigma itself
sends any index that is not equal to i1, i2, …, ik to itself , so j is
not in i1 up to ik, sigma of j is j, so sigma k of course will also
send j to j, so in other words sigma k sends any index to itself. So
sigma k is identity and sigma any smaller power cannot be
identity, so this proves the proposition.

So sigma which is a k-cycle has order k, and so order of a k-cycle

is k.
(Refer Slide Time: 25:44)

So for example the order of (1 2 3 5) is 4 because it is a 4-cycle.

And it is a good exercise for you to check that in fact it is 4 by
multiplying it out. So maybe I will just do quickly (1 2 3 5)
squared, is 1 3, 3 goes to 1, so that is (1 3) . And 2 goes to 3 and 3
goes to 5 so (2 5). What is (1 2 3 5) cubed? So 1 goes to 5 and 5
goes to 3, and 3 goes to 2 and (1 2 3 5) to the 4th is identity. So
this I will let you check, so check this, and of course the point of
previous proposition is that you don’t have to do this calculation
every time. (1 2 3 5) you can immediately say, because it is a 4-
cycle, it has order 4.
Online Editing and Post Production

Karthick

Ravichandaran

Mohanarangan

Sribalaji

269
 Komathi

Vignesh

Mahesh kumar

Web Studio Team

Anitha

Bharathi

Catherine

Clifford

Deepthi

Dhivya

Divya

Gayathri

Gokulsekar

Halid

Hemavathy

Jagadeeshwaran

Jayanthi

Kamala

270
Lakshimipriya

Libin

Madhu

Maria Neeta

Mohana

Mohana Sundari

Muralikrishnan

Nivetha

Parkavi

Poonkuzhale

Poornika

Premkumar

Ragavi

Raja

Renuka

Saravanan

Sathya

Shirley

271
 Sorna

Subash

Suriyaprakash

Vinothini

Executive Producer

Kannan Krishnamurty

NPTEL Co-ordinates

Prof. Andrew Thangaraj

Prof. Prathap Haridoss

IIT Madras Production

Funded by

Department of Higher Education

Ministry of Human Resource Development

Government of India

HYPERLINK "http://www.mptel.ac.in"www.mptel.ac.in

Copyrights Reserved

272
NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract Group Theory
Module 05
Lecture 28 –“Symmetric groups III”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE
(Refer Slide Time: 00:25)
But now what about products of disjoint cycles, what about the
order of product of disjoint cycles? And this is dealt by next
proposition. It says that if sigma is an element of symmetric
group has cycle decomposition, let us say sigma equals sigma 1,
sigma 2 up to sigma k. So remember cycle decomposition
always assumes that the cycles are disjoint cycles.
And let us say sigma 1, sigma i is a mi cycle. So in sigma 1 is a
cycle of length m1, sigma 2 is a cycle of length m2, sigma k is a
length of mk. Then order of sigma is lcm, least common
multiple of m1, m2, mk.
(Refer Slide Time: 01:57)
So proof, so first of all, LCM stands for least common multiple.
So now we have in the previous proposition showed that if you
have a single cycle its order is the length of that cycle, so if it is
a k-cycle its order is k. Now I am, in the new proposition, I am
telling you if you have a cycle decomposition into a product of k
cycle, which are disjoint, that is important, and then the order of
the product is the lcm of the individual orders. Remember order
of sigma i is mi, and the order of sigma is lcm of mi. So this is
very easy to prove.

273
So by the previous proposition, let us quickly prove this, by the
previous proposition, order of sigma i is mi, let us keep this in
mind, for i from 1 to k. So order of sigma is mi, because sigma i
is an mi cycle. So it has order mi.
So let M be the lcm ,so for simplicity let capital M be the lcm of
m1 to mk. Now what is sigma power m. Let us compute sigma
power m. This is sigma 1, sigma 2, up to sigma k power M. So
remember this means I am doing sigma 1, sigma k, sigma1,
sigma k, sigma1, sigma k M times. So in general permutations
do not commute with each other, but disjoint cycles do. So you
can sigma 1, sigma k is sigma k, sigma 1. Because they are
disjoint cycles.
Because they are disjoint cycles, we can interchange them, and
in a group we can apply associativity law, so we can remove the
brackets first. Sigma k and sigma 1 can be interchanged. And
then you reorganize all of them and put all sigma 1 to the
beginning, so you get sigma 1 power M. Then you put all sigma
2s , basically commutativity of these sigma i means that you can
arrange them in any order.
I will put all sigma 1s, there are M of them, then all sigma 2s,
then all sigma 3s, and finally all sigma ks, so I can write this
sigma power M as this. But now M is the LCM of mi, small mi,
so order of sigma 1 is m1, and m1 divides capital M, because
capital M is the LCM of m1 through mk, so in particular m1
divides M. This implies, sigma 1 power M is identity. This is
something we have seen in detail in various situations, if you
have order times something, power of that will be identity.
Similarly, sigma 2 power M is also identity because m2 also

274
divides M, m3 divides M so that is e, so everything is e. So
sigma power M is e. This means, order of, remember order of
sigma is supposed to be M. So we have to prove that order of
sigma is M. So in other words, the least positive integer d such
that sigma power d is identity, is M. We have checked that sigma
power M is identity. We now need to check that nothing smaller
than M can make identity.
(Refer Slide Time: 06:19)
So suppose sigma power n is identity, by the same calculation
above, so e is sigma power n, just like we calculated here,
because sigma 1, sigma2, sigma k commute with each other, I
can write sigma power n as sigma 1 power n , sigma 2 power n ,
sigma k power n. I can re-arrange them, so that all sigma1s
comes first, sigma 2s come next, sigma 3s come next and finally
sigma ks come next.
Now let us stare at this equation for a while, remember sigma 1,
sigma 2, sigma k are disjoint cycles, so whatever index appears
in sigma 1, does not appear in sigma2, does not appear in
sigma3. But after applying, suppose i is an index appearing in
sigma1, then so i does not appear in sigma 2, sigma 3, because
sigma 1 is disjoint with sigma 2, i appears in sigma1, so it does
not appear in sigma2 and doesn’t appear in sigma3 and sigma k.

Now I claim that this implies i does not appear in sigma 2 power
n also, sigma 3 power n also, sigma k power n also. Because
remember sigma 2, when you apply to itself, it only talks about,
deals with indices that originally appear sigma 2. In the previous
examples when we proved to previous proposition, remember if

275
we take (1, 2, 3, 5) as a cycle, its products with itself only
involve the original indices that appear in sigma, in this example
it is 1,2,3,5. So products will only involve 1,2,3 5.
Suddenly a new index cannot appear right, because sigma fixes
an index powers of sigma continue to fix that index. So we don’t
have to worry about if i does not appear in sigma 2, i does not
appear in sigma 2 squared. Sigma 2 power 3 , sigma 2 power n.
So i only appear in, hence i only appears in sigma 1 power n, so
i only appears in sigma 1 power n. May be it doesn’t appear in
that also, because may be it is fixed by I, but it cannot appear in
sigma 2 power n and sigma k power n.
So now e fixes i that because e is the identity element, so sigma
n fixes i, because sigma power n is e, I am assuming sigma
power n is e for some n. Sigma power n fixes i, this implies
sigma power 1, sigma 1 power n, sigma 2 power n, sigma k
power n fixes i. We just argued in the right hand side of this
slide that sigma 2 power n fixes i, it does not appear in i means, i
doesn’t appear in sigma 2 power n means it fixes i. Sigma k
power n also fixes i, so we can conclude that only, so it fixes i,
so sigma 1 power n fixes i, so sigma 1 power n must fix i, it
cannot, see if it sends i to something else, because i is not
present in sigma 2 and sigma k, you cannot send i back to itself
under this product.
So sigma 1 power n fixes i. Similarly sigma 1 power n fixes
every index appearing in sigma 1. So i was an index appearing
sigma 1, if I was an index appear in sigma 1 , because sigma 2
power n, sigma 3 power n, and sigma k power n, they all fix ,
we can be sure that sigma 1 power n also fixes i. Because if doe

276
not fix i the product cannot fix i, that is the point, because if i
goes to j, how will j come back to i, under the product is suppose
to come back to i, under the product of sigma 1 power n and
sigma 2 power n , sigma k power n , i goes to i, so if i goes to j
under sigma 1 power n j must come back to i, but it cannot,
because if i doesn’t appear in these indices, j also wont appear in
these indices.

So i must go to i under sigma1 power n, similarly everything

that appears in sigma 1 must be fixed by sigma1, this is to say,
sigma 1 power n is identity. So sigma 1 power n must fix
everything that is in sigma 1, of course it fixes everything that is
not in sigma 1. Because sigma 1 fixes everything that is not in
sigma 1, so sigma 1 power n is identity. It fixes every index,
(Refer Slide Time: 12:16)
but then order of sigma 1 is m1, and sigma 1 power n is identity.
These two together imply that m1 divides n. This is also an
exercise I have done in the previous videos, if order of an
element is m but some other power of that element is e, then
order must divide n. We have argued with sigma 1, we can argue
with sigma 2, the same argument will show that, sigma 2 power
n is also identity.

There is nothing special about sigma 1, sigma 1 power n is

identity means sigma 2 power n is also identity. So m2 divides
n , and so on, order of sigma k is mk and sigma k power n is also
identity. By the same argument mk divides, so each mi divides
n. Since each mi divides n, for all i , we have the lcm divides n.
See if each mi divides n , n is a multiple, n is a common multiple

277
of mi, but capital M is the least common multiple, and the
property of least common multiple says that, least common
multiple divides all multiples.

And in fact definitely we have, even if you don’t agree this or

you are not familiar with this it is the least common multiple, M
is the least common multiple and n is a multiple, so we must
have this, M is the least, we only need this for the moment,
capital M is the least common multiple, n is a multiple, so M is
less than or equal to n. So what we have shown is that now
coming back to our sigma, sigma power M is identity and if
sigma power n is identity for a positive integer, then m is less
than n. So together these two facts imply that order of sigma is
M. So this proves the proposition.
(Refer Slide Time: 15:07)

What we have shown is that if you have a product of disjoint

cycles then order is the common multiple of the sizes, least
common multiple of the orders of each individuals disjoint
cycle.
So for example if you take the element we started with in the
video (1 3 2 4) (5 6) (7 8) . So this has order 4, order 2, order 2.
So what is the order of sigma? Recall that these are disjoint
cycles which is important, so order is lcm of 4, 2, 2 which is 4.
So order of sigma is 4.
So as I said it is important that the cycles have disjoint
decomposition, the cycles are disjoint. Let us look at another
example, let us take sigma to be (12)(123), these are not disjoint

278
cycles. What is the order of this? In order to find that let us
multiply it out. Where does this go? This we have done before, 1
goes to 2 , 2 goes to 1, so 1 is sent to itself, 2 goes to 3, and 3
goes to 1 and 1 goes to 2, so this is (2 3).
So sigma is a 2-cycle, but it is a product of, it is also a product of
a 2-cycle and a 3-cycle. Sigma is a 2-cycle but it is also product
of a 2-cycle (1 2) and 3-cycle (1 2 3). If the proposition applied
to any product, order of sigma, actually it is two, because it is a
2-cycle. And we have already proved in a previous proposition
that k cycle has order k, so order of sigma of 2, but lcm of 2 and
3 is 6. 6 is different from 2, so the proposition requires that the
cycles are here.
(Refer Slide Time: 17:47)

The proposition does not apply, because (1 2) and (1 2 3) are not

disjoint cycles, so to find the order of a permutation we have to
find its disjoint cycle decomposition, if you have some cycle
decomposition, you cannot say order is lcm of the individual
orders. So only if you have disjoint cycle decomposition you
have the property that order is the lcm of orders.

So I am going to stop the video at this point. In this video we

have looked at cycle notation for elements of symmetric group,
we have seen that every element of symmetric group has a
decomposition into disjoint product of cycles or a product of
disjoint cycles, we have seen that order of a k-cycle is k and we
have seen that if a permutation has a decomposition into disjoint
cycles, order of the permutation is lcm of the individual orders.

279
In the next video we are going to further study the cycle
decomposition. Thank you.
Online Editing and Post Production
Karthick
Ravichandaran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh kumar
Web Studio Team
Anitha
Bharathi
Catherine
Clifford
Deepthi
Dhivya
Divya
Gayathri
Gokulsekar
Halid
Hemavathy
Jagadeeshwaran
Jayanthi
Kamala
Lakshimipriya
Libin
Madhu
Maria Neeta
Mohana

280
 Mohana Sundari
Muralikrishnan
Nivetha
Parkavi
Poonkuzhale
Poornika
Premkumar
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirley
Sorna
Subash
Suriyaprakash
Vinothini
Executive Producer
Kannan Krishnamurty
NPTEL Co-ordinates
Prof. Andrew Thangaraj
Prof. Prathap Haridoss
IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resource Development
Government of India
www.mptel.ac.in
Copyrights Reserved

281
 NPTEL

NPTEL ONLINE COURSE

Introduction to Abstract

Group Theory

Module 06

Lecture 29 – “Symmetric groups IV”

PROF.KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

So we have seen so far, that any permutation is SN can be written

as a disjoint product of cycles, and that is very useful, in
particular, for example, you want to calculate the order of the
permutation, because the order of permutation, will simply be as
LCM of the orders, provided the cycles are disjoint. Next we are
going to look at another decomposition for permutations, which is
not disjoint, but it is useful in some other situations.

(Refer Slide Time: 00:44)

So let us define a “transposition”, a 2-cycle is called a

“transposition”, so 2-cycles are particularly simple permutations
right, what is a 2-cycle? It is something like,(ij), right (ij) is a 2-
cycle. What does it do as a permutation, it sends i to j, and j to i,
and fixes everything else, and so permutations which are 2-cycles
are easy, they just change from i to j, and j to i, and fix everything
else, they are called transpositions, because they are just

282
transposing Iiand j, so you are putting i in place of j, and j in place
of i, we transpose them, and we do not disturb the other indices.

So the transpositions are easy to understand, and a transposition

has order 2, the transposition has order 2 because it is a 2-cycle, so
2-cycles have order 2, in general k cycles have order k.

Now let us see, if we can write any element, let me write the
proposition, and then we will look at some examples, and then
prove that.

(Refer Slide Time: 02:21)

Every permutation can be written as a product of transpositions or

2-cycles, but, the most important thing to remember is, not
necessarily disjoint transpositions, so we can write it as a product
of transpositions, but in order to achieve this, we have to give up
disjointness, we cannot write disjoint as disjoint transpositions, so
before I will proving this, before proving this which is actually not
difficult at all, let us look at some examples.

So let us take σ to be (132) in S5, so actually it can be S3,S4,S5,S6

and so as on, what does it do, it sends 1 to 3, and 3 to 2, 2 to 1, so
it is a 3 cycle.

So remember that it cannot be written as a disjoint, we certainly

know, σ cannot be written as a disjoint transpositions, so let me
write this cannot it be written as a product of disjoint
transpositions. Why can’t it be written as product of disjoint

283
transposition, suppose it can be written so if σ is σ 1, σ2, σ k,, where
σ1 to σ k are disjoint transpositions. So the reason for why we
cannot write σ as a product of disjoint transpositions. If you can
then the order of σ is lcm (order (σ 1) …. order(σ k)), and because it
is a product of disjoint cycles, but order of σ1 is 2, σ2 is 2 , σ k is 2,
because there all are transpositions.

So the order of σ is 2, the lcm of 2,2,2 any number of times, is 2,

but σ is a 3-cycle, so order of σ is 3, by another proposition that
we have proved sometime ago, order of σ is 3, because it is a 3-
cycle, so if it can be written as a order of product disjoint 2-cycles,
its order would be 2, whereas its order is 3, so cannot be written as
product of disjoint transpositions. However, we can write it as
product of transpositions, if we do not insist on disjointness, how
do we do this?

(Refer Slide Time: 06:17)

So (132), for example is it is easy to check, (12)(13), why is this?

Because, you can check quickly, this sends 1 to 3, 3 to 2, and 2 to
1, what does this do, 1 to 3, under the 1 st one, but then 3 to 3,
under the 2nd one, which is good, where does 2 goes 2 , under the
1st one, and 2 goes to 2 , under the 2 nd one, which is also good.
Finally 3 goes to 1, and 1 goes to 2, so these are equal right, so
(132) is written as a product of two transpositions, but they are not
disjoint because, 1 is common to both, so they are not disjoint, as
we know, you cannot write it as a product of disjoint
transpositions.

284
I did not mention this earlier because we do not need it, but when
you write a permutation as a product of disjoint cycles, that
product or decomposition is unique, it is not at all difficult to show
that, because they are disjoint cycles, you cannot have another
product decomposition to disjoint cycles. However because we
cannot have disjoint product of transpositions, this product into
transpositions is not unique, okay. So (132) happens to be (13)
(32), as you can quickly check.

What happens to 1 in this, 1 goes to 1 in the 1 st one, 1 goes to 3

under the 2nd one, 2 goes to 3 under the 1st one, 3 goes to 1, under
the 2nd one, 3 goes to 2, under the 1 st one,2 goes to 2 under the 2nd
one, so this is same as (132), right we have proved this. So this is
a different decomposition, of the same element, same permutation,
namely (132), as a product of transposition, okay. So certainly it is
different because, it is (12) and (13) come here, (13) and (32), so
they are different, transposition appearing in this, so these are
different decomposition, but again it is not disjoint, because 3 is
common in both.

So you might wonder, may be the decomposition is not unique,

but the number in which, number of transposition is unique, here
there are 2 transpositions, here also there are 2 transpositions but
as another example, we will show, that even that number is not
unique.

(Refer Slide Time: 09:27)

285
So let us say, I take σ to be (12)(13)(14), so this is a product of 3
transpositions, but you can quickly check that, this is same as, I
have checked this, you can do this as an exercise, (23)(25)(12)(45)
(15), okay for example, let us randomly pick some element, what
happen to let say 1 under both of this. 1 under this decomposition,
goes to 4, then 4 goes to 4, and 4 goes to 4, so 1 goes to 4 under
the 1st one, what happens to 1 under the 2nd one,1 goes to 5, 5 goes
to 4, and 4 is not appearing there, so this is 1 goes to 4, under the
same, so similarly you can check that, these two decompositions
are the same.

So here you need 5 transpositions, so not only is the

decomposition into product of transpositions is not unique, here
the number of them is not unique, so we need 3, and we need 5,
however, there is some uniqueness in this decomposition into a
product of transpositions that I will come to you later, now let us
come back to the proof of the proposition.

What do we have to show, we have to show that given any,

permutation, in Sn, it can be written as a product of transpositions,
first observe that, it suffices to show that a cycle σ can be written
as a product of transpositions.

Why is this? So suppose I show that a cycle can be written as a

product of transpositions, then why am I able to say that any
permutations can be written as a product of transpositions?
Because given any permutation, given any permutation σ , we can
write σ as σ1σ2…σ k, where σ1 to σ k are what, disjoint cycles, here
disjointness is not that important, but we can write it as a product

286
of cycles, that we already know, we have proved this, any
permutation, in S n, it can be written as a product of cycles, now
suppose if I prove that cycles can be written as a product of
transpositions.

(Refer Slide Time: 12:59)

So σ1 can be written as a product of, so we have σ 1, σ2, σ k, but σ1

can be written as a product of transpositions, σ2 can be written as a
product of transpositions, so this is σ1, this is σ2 and so on, σ k can
be written as a product of transpositions, then we have written σ
right remember we are not claiming anything about disjointness,
in fact we cannot claim, there cannot be a decomposition into a
product of disjoint transpositions, so we do not care if repeated
things happen, σ1 can be written as a product of transposition so
can σ 2 and so can σ k. So σ can also be written as a product of
transpositions.

So we assume now, σ is a let’s say k-cycle, so we only need not

consider the case of k-cycles, so say σ = (i1,i2…ik ).

So now if you just stare at this for a bit, you can see how to write
this as a product of transpositions. So we can write, σ as so I am
going to write it like this, so keep in mind that what σ does is
sends i1 to i2, so let me put (i1, i2) here. I am going to write from
left to right, but when we actually multiply remember, to figure
out the image of an index we go from right to left, so if this is the
last one, that is there, and if I make sure that i1 does not appear in
the remaining ones, then i1 will go to i2 in under the product, so I
will do (i2,i3), so now whatever I write here let’s ensure that i2
does not appear in any of these, then in order to figure out, what is

287
the image i2 under this product, we have to keep going from right,
and we hit i2 here, so i2 will go to i3 under this, but i3 is not
present in this, so i2 will go to i3, which is how it should be.

So now I will put (i3, i4), and I will keep doing this, (i(k-3),i(k-
2)),(i(k-2),i(k-1)),(i(k-1),ik), okay. So I claim this, we can write
like this, in fact we claim that, this is an equality, so let us check
this as I said, it is very easy to see this, but let us check one by
one, where does i1 go?

For the right hand side product, for the product on the right hand
side, i1, remember to figure out where i1 goes we have to go from
right, and keep applying i1 every time, we see.

So i1 is not present here, it fixes i1, so we do not care, i1 is not

present here, i1 is not present in the 3 rd one, from the right, it is not
present in any of the 1st k-1, it is present only in 1 st k-2, it is
present only in the last 1, so we must send i1 to i2, right, σ also
sends i1 to i2 remember. Now where does i2 go, we have to go
from right and see where i2 first appears, and the 1 st time i2
appears remember, the way we are writing this, i2 will not appear,
until we hit (i2, i3).

So i2 will go to i3, right, is that clear, because i2 is not present, i2

is already written here in (i2, i3), so subsequent transpositions will
only involve higher indices, so i2 will not appear there, i2 will go
to i3 here, now possibly, we have to go left and see, but i3 does
not appear in the previous one, so i2 will go to i3 under the
product. What happens to i3, i3 will not appear, until we hit (i3,
i4), and here it will go to i4, but then i4 no longer appears again, i4
does not appear here, so i3 goes to i4, and so on.

288
What happens to i(k-2)? i(k-2) goes to i(k-1), the 1 st one on the
right does not involve i(k-2), the 2nd one does, so i(k-2) goes to
i(k-1), but i(k-1) does not appear again anywhere else, so i(k-2)
goes to i(k-1), the point of this decomposition is, each index
appears exactly twice, except i1 and ik, i2 appears here twice, i3
apears twice, and i(k-1) appears twice, but once i(k-2) appears
here, only time i(k-1) next it is in the right hand side permutation,
so it does not matter what happens previously, i(k-2) goes to i(k-
1).

So the 1st one itself involves i(k-1), so we have to check ik, but
then ik never appears again anywhere, so i(k-2) goes to i(k-1), so
finally check ik. So ik goes to i(k-1) on the 1 st thing, now i(k-1)
appears again, so it sends i(k-1) to i(k-2), i(k-2) goes to i(k-3)
here, and you have to keep tracing back, i(k-3) goes to i(k-4), i4
goes to i3, i3 goes to i2, i2 goes to i1, so in order to determine the
image of ik, we start with the right most permutation, and keep
tracing it back, all the way back to i1, so ik goes to i1, this is
exactly equal to σ, this is exactly equal to σ.

(Refer Slide Time: 19:50)

So we have shown that σ is (i1, i2)(i2,i3)(i3,i4)….(i(k-3),i(k-2))

(i(k-2),i(k-1))(i(k-1),ik), and this is a product of transpositions, so
we have shown that, give me any k-cycle, (i1, i2….ik), I can write
it as a product of transpositions, and we have already said that any
permutation is a product of cycles, and every cycle can be written
as a product of transpositions, so any permutation can be written
as a product of transpositions. This proves the proposition, and we
have already looked at some examples.

289
So the important points to keep in mind is, first, you can write any
permutation as a product of transposition. This product is not
disjoint, and not unique, even the number of transpositions
required is not unique, and in the next video, I am going to look at
what kind of uniqueness we can get out of this decomposition into
product of transpositions, thank you.

Online Editing and Post Production

Karthik

Ravichandran

Mohanarangan

Sribalaji

Komathi

Vignesh

Mahesh Kumar

Web Studio Team

Anitha

Bharathi

Clifford

Deepthi

Dhivya

290
 Divya

Gayathri

Gokulsekar

Halid

Heamvathy

Jagadeeshwaran

Jayanthi

Kamala

Lakshmipriya

Libin

Madhu

Maria Neeta

Mohana

Mohana Sundari

Muralikrishnan

Nivetha

Parkavi

Poonkuzhale

291
 Poornika

Premkumar

Ragavi

Raja

Renuka

Saravanan

Sathya

Shirley

Subash

Suriyaprakash

Vinothini

Executive producer

Kannan Krishnamurty

NPTEL Co-ordinators

Prof.Andrew Thangaraj

Prof. Parthap Haridoss

IIT Madras Production

Funded by

292
 Department of Higher Education

Ministry of Human Resource Development

Government of India

HYPERLINK "http://www.nptel.ac.in" www.nptel.ac.in

Copyright Reserved

293
 NPTEL

NPTEL ONLINE COURSE

Introduction to Abstract

Group Theory

Module 06

Lecture 30– “Odd and even permutations I”

PROF. KRISHNA HANUMANTHU

CHENNAI MATHEMATICAL INSTITUTE

In the last couple of videos we looked at the symmetric group on

N letters, we saw how to represent permutations which are
elements of SN as cycles and we learned that any permutation can
be written as a product of disjoint cycles and we also saw that any
permutation can be written as a product of transpositions which
may not be disjoint, we saw that a cycle of length K has order K
and if we have a product of disjoint cycles the order of the product
is simply the lcm of the individual orders. So these are the various
things that we have done in the last few videos. I want to now
spend more time on writing a permutation as a product of
transpositions.

(Refer Slide Time: 01:04)

So if you recall one of the example from, from the previous video
we wrote the permutation (12)(13)(14) and we saw that this is

294
same as (23)(25)(12)(45)(15), okay, so this is the same
permutation but on the left hand side we have a product
description which has which requires 3 transpositions on the right
side we have 5 transpositions okay, I, I remarked on this last time,
the description product of transpositions is not unique and even
the number of transpositions that is required that is also not
unique, we have 3 here we have 5 here.

However what is unique is whether we need an odd number of

permutations or transpositions are even number of transpositions,
so this is one of the examples. We saw another example, we saw
that if you take the 3-cycle (132) this can be written has (12)(13),
this can also written as (13) (32), okay so here we need 2
transpositions, in this also we need 2 transpositions, both even, of
course in this case both 2. In this case these are both odd, even
though it is 3 and 5.

So the theorem that I want to prove in this video and it will take
me some time to prove this, we will do a various examples to
understand the proof and then I will prove. The theorem is
informally if you want a theorem it says that if permutation as a
description of the product of transpositions in 2 different ways the
number of permutations the number of transpositions required in
both the things have same parity, so they are both even or they are
both odd.

295
So let me write this more precisely, mathematically, how do we
write this? Let me write this as fallows. Let ρ be a permutation, so
I am going to use a new Greek letter because I am going to use tau
and σ to denote transpositions, so you read this as a rho, rho, okay
so let this be any permutation.

Suppose it a 2 representations. Suppose that rho is on the one hand

it is σ 1σ 2… σ k, on other hand it is also same as tau 1 , tau 2 , tau
t, okay where these are all transpositions, σ1 through σ k and tau 1
through tau t are 2-cycles, remember transpositions are just
another name for 2-cycles. So what is the situation? We have a
fixed permutation rho, and we have written in fashion as σ 1 to σ
k and in another fashion as tau 1 to tau t. Certainly we cannot say
k = t but what we can say is that, then both k and tau, sorry t, k is
the number of σs and t is the number of tau-s, okay so let me
remove the word both here. So then k and t are both even or both
odd, so in other words, what I am saying is that they one of them
they cannot be one can’t be even and the other odd, that is not
possible, either both are even or both are odd as illustrated in this
example, you have 3 and 5 both are odd 2 and 2 both are odd
okay. So you can have both odd or both even.

So the rest of the video is focused on proving this theorem. It will

take some set-up, so I am going to slowly introduce to you the
techniques involved, so this is not so easy to prove though it
would be, as we proceed with the proof, hopefully you will follow
the proof and it makes sense to you but the proof is somewhat
lengthy.

(Refer Slide Time: 05:50)

296
So let us systematically prove this. The goal of this video is to
prove this theorem, okay, so I will systematically and slowly go
through the proof. So in order to prove this let us introduce a
polynomial, so let us introduce a polynomial okay, so if you are
not very familiar with the notion of polynomial, do not worry
about it, we do not need a lot about them, it is just a formal thing
and I will do enough examples to make it clear.

So I am go to call this polynomial f x1 up to xn, so think of x1

through xn as variables, are variables or indeterminates. So think
of them as just meaningless symbols, okay, they do not have any
meaning, they are just symbols, so I will define this to be the
following. So I will start with x1, I will subtract from it x2, then I
will subtract from it x3, x1-x2, x1-x3, I will do it all the way up to
xn, okay so x1 minus the rest, one by one, then I will start with x2
okay and I will x2 to the right of x2 there is x3 so I will start with
x3, x2 minus, okay so to basically what we have is x2-x3 , x2-x4
and up to x2- xn, all the way and I finally have x(n-1)–xn , okay
so this is the polynomial and formally, if you write compactly if
you want to write this this is written like this.

(Refer Slide Time: 07:53)

We take this disjoint, product this symbol stands for product, you
are all familiar with the summation sign, this is just the product

297
sign, we take i and j distinct, i<j, i can be anything from 1 to n-1
and j can be anything from 2 to n right. i can’t n be because i in n
there will be no j, j must be strictly more than n, similarly j cannot
be 2 because j must be strictly more than 1 so this is the product.
For any pair of indices where i is less than j, you introduce a term
xi-xj oaky.

So and remember n is fixed in the theorem. We are working with

the symmetric group of n letters, n is any positive integer, we do
not want to work with any specific one, the theorem works with
any positive integer. But as an example what would be x1, x2 so
f(x1) is just, that does not exist, there is no, we don’t have two
indices like that if you have n =1. So f(x1, x2) would be simply
x1-x2. There is just one pair of indices,1 and 2.

What would be f(x1,x2,x3)? It would be, you take a x1-x2 , x1-x3,

so starting with x1 you take every other index, 1 is fixed so 2 and
3 and then starting with x2 we take every other index and is there
is anything else? No, we have you have , here you have i =1 j=2
but the only possibilities here you have i= 1 in this case j=2 or 3
and i=2 you have j=3 okay these are the 3 possibilities there will
be 3 terms in the product. So this is x1-x2 times x1-x3 times x2-
x3.

298
Similarly I will write one more and hopefully it will become clear
to you what is the polynomial. It is a, is an expression, think of it
as an expression involving variables x1,x2, …, xn. So here for i=1
you have for i=1 you have j =2, 3 or 4, right so for i=1 you have
j=2,3 or 4, so you have x1 –x2, x1-x3 ,x1-x4, okay so what about
i=2? So you have i=2, j must be bigger than 2, so j is 3 or 4, so
you have x2-x3 x2-x4, and then i=3 then j must be 4, so you have
x3 –x4 right. So these are the 6 terms, so this polynomial for
x1,x2,x3,x4 will be these 6 terms and so on. Now we can think of
what x1,x2 f(x1,x2,x3,x4,x5) would be it will be x1-x2, x1-x3, x1-
x4, x1-x5 and so on. So this is what f is.

(Refer Slide Time: 11:47)

So now what I want do is I want to take an element sigma in Sn,

okay so σ is just a permutation on n letters and I want to define σ*
of f as fallows. So σ * of f will be, so f is remember this,
sometimes I do not write the variables, so we have so f is the
product xi –xj, 1<i and i strictly < j and j <= n, so σ *f, I will
define, so I am going to modify f so σ *f is a modification, I am
going to change f a little bit, of f using σ okay, so think of it like
this, σ *f is simply a modification using σ. How do we modify?
Remember what is σ? Sigma just is a permutation of the letters.

So what I will do this, I will continue to take the product over it

pair i, j i strictly < j, but instead of xi-xj, I will have x σ i – x σ j,
oaky. This might seem somewhat mysterious but it is not, so I just
want to make it very clear to you because it is important to
understand what σ * f is before we proceed.

299
So what is σ *f? I am going to start with f and modify indices
based on σ. So as an example, okay so let us suppose we take 2
variables so then so take n=2, so then f is simply x1-x2. So if you
take now we are dealing with S2 right so n is 2, S2 is just 2
elements e and (12). This we have seen, the order of S2 is 2
factorial, right and which is 2, it is e and (12) so if you take for
any σ we can define, so let us compute this in these two cases.

(Refer Slide Time: 14:34)

e * f is this is a compact way of writing, xσ i –x σ j, but if you

spread it out what would it be? e is identity element so nothing
changes so you have x (e of 1) –x (e of 2), but e of 1 is 1, e of 2 is
2. So this is just x1-x2, which is just f, right.

On the other hand, what is 12*f? So this is x okay, so now it is

easier for me to use the letter σ, so then this is x (σ of 1) – x (σ of
2), so when we right σ of i, remember σ is a function from 1
through n to 1 through n. So σ of i is simply the image of i, okay
under σ. So when I write σ of i, I am actually thinking of sigma as
a function, so what is what is image of 1 under σ which is (12), it
is 2 so it is x2 –x σ 2 will be x1.

So this is not same as f right, this is actually –f. So here we have

changed the sign, x1-x2 becomes x2-x1. So this is what happens
when n=2.

300
Let us do n =3 okay. So here what is f? Here f is x1-x2, x1-x3,x2-
x3 okay, and here S3 remember has 6 elements, so they are e, (12),
(13), (13),(23),(123),(132) okay. Let me do not all of them but
illustrate the point, let me do σ = (13) for example.

What is σ *f? So you start with f and you keep the order but you
just change the subscripts and see what happens. So it would
become x σ 1 – x σ 2, so I am looking it f here x1- x2 that
becomes this, x1-x3 becomes x σ 1 – x σ 3, and finally we have x
σ 2 –x σ 3. So σ *f is (x σ 1-x σ 2) (x σ 1 - x σ 3) (x σ 2 – x σ 3).

But now let’s actually compute these things. x is σ 1 remember is

σ is (13) so this is x3. σ 2 what does σ do to 2, nothing so it is 2, 2
is fixed by σ so x3 – x2. σ 1 is 3 so this is x3 – σ 3 is 1 so this is
x1, σ 2 is 2 so this is x2 minus σ 3 is 1, so this is x1.

So σ * is (x3 –x2)(x3-x1)(x2-x1) and what is this? This is how do

we compare this with f, so every term remember here is switched
x1- x2 became x2-x1, x1-x3 became x3-x1 , x2-x3 you have a x3
–x2, so you have -,-, -. So this is –f. Is that clear σ f is – f, because
f is a product of 3 terms, σ *f became, those 3 terms are
interchanged, the order in which the 2 terms, x1-x2 is
interchanged this is – okay so this is –(x2-x3) –(x1-x3) , this is -
(x1-x2) right and so this is – f , -1 time -1 times -1 is -1, and then
you have just f.

(Refer Slide Time: 19:26)

301
So σ *f is f. Let us do one more example let us say T = (123), that
is another element of S3, so Τ is 123. So what is Τ *f same thing
so f is, the same process, it will be different but the same
procedure we must follow. f is x1-x2, x1-x3,x2-x3 so Τ*f will be x
(Τ1) –x (Τ 2 ) x(Τ1)- x(Τ3) x(Τ2) – x(Τ3) right so this should be Τ
1 is 2, Τ 2 is 3, x2-x3, Τ1 is 2, Τ3 is 1 so this is x2-x1, Τ2 is 3 and
Τ3 is 1, okay so this is x3-x1.

So now how do you compare this with f, f has 3 terms again

remember that f has 3 terms x1-x2 ,x1-x3,x2-x3 here we have
those three terms but unlike in the previous example one of those
terms has not changed x2-x3 remains x2-x3 but other two things
we have interchanged. This and this have changed size , right,
because it was x1-x2 in f, that has become x2-x1, it was x1-x3 in f
that has become x3- x1, x2-x3 remains as it is, so this is actually
just f because this is minus of that term times minus of that term,
-1 times -1 is 1 so Τ *f is f okay.

(Refer Slide Time: 21:31)

So what I want to now say is that and I hope you agree based on
these examples, so now I am going back to the general situation,
if σ is a permutation in SN then σ * of f is either f or –f okay. That
is what we found in these examples right it becomes either f or –f.
Of course doing examples is no a proof of this, the examples are
supposed to give you an idea of why it must be true and how to
prove it. And what is a proof?

302
So I won’t write this in detail because it becomes, it is easier if
you think about it yourself and try to write down a proof, but I
will give you a basic reason why this must be true. Because
remember f was product xi –xj okay, so it is over all i,j and σ *f is
the product x σ i –x σ j, over the same i and j. So if you think
about it because σ is a permutation σ is a bijection right remember
symmetric group is the set of bijections of this set.

Because it is a bijection, as i varies over 1 to n or 1 to n-1, in this

case, σ i also varies, similarly σ j also is an element of 1 to n. So if
you take an individual term in f that is translated to some other
term only changes will be negative signs as we have seen in this
example. For example, x1-x3 became x3-x2 so we, we must allow
that in the definition of f we have xi-xj were i is strictly < j
always, so 12,13,23 right 12,13,23, in f you never have a bigger
index minus a smaller index.

f is defined to be x i- xj where i is strictly <j, but once you do σ *f,

that will no longer be the case. Sometimes we have a bigger index
minus smaller index: x2-x. Because if i<j certainly it is not true
that σ i is < σ j, that is not the case. But however that is just a
negative of, if you have a bigger index – smallest index, and that
is negative of smaller index – bigger index, which appears in f, so
f is a product of some terms, σ *f will be product of same number
of terms.

But some of these terms, it will be the same terms with possibly
negative sign, so if you now accumulate all these negative signs
you might either have + or –, as this example shows, you have 3

303
negative signs giving you –f when σ was (13), when Τ is (123)
you have only 2 negative signs, so that gave you f. So σ *f must
be either f or –f, the same terms are preserved up to a change of
sign, so σ * is either f or –f. Let me say this is my proof. I won’t
go into a more formal proof, but I hope this is clear enough for
you . So σ *f is f or –f.

(Refer Slide Time: 25:34)

Next I want to do, if σ and Τ are in SN, what is σ Τ *f? This is my

next question. σ Τ is another permutation right, so I can apply * of
it. How is it related to σ *f and Τ *f, okay. I want to understand
how is it related to this. So again I will explain this by the example
that we did. So take n = 3 and σ was 1 let’s see σ was 13 and Τ
was 123. So what is σ Τ, so σ Τ we are now okay we are all
become comfortable hopefully with products of permutations,
cycle products.

So here σ Τ will be 1goes to 2, and 2 is not, 2 is fixed by sigma, so

that’s 1 goes to 2, 2 goes to 3, 3 goes to 1, so that is 12 so you
close, 2 ->3 3->1 so 12 is a cycle 3->1, 1->2 so σ Τ is (12). And
what is σ Τ *f? If you now do the calculation that we have done
earlier, it is x, so f was remember I will write it again here for
convenience, f is x1-x2 , x1-x3 ,x2-x3. So σ Τ *f will be I will just
directly write is 1 ->2 in this so x2 and 2->1 so that is x1- x2, 1 ->
2 so x2 and 3 is fixed so that x2 –x3 and 2->1. So and 3 is fixed so
x2-x1 right.

304
 I will just look at the subscripts which are the indices of variables
and see where they go under σ Τ. So 1->2, and

2->1, so 1st one changes sign and the 2nd one 1->2 , 3->3 so x2-x3,
this does not change sign, no change of sign. Here 2->1 and 3->3,
so here also no change in sign, right in the 1 st one change in sign
okay. So what is this so this? This is –f right, because there is one
change of sign. So it is minus of x1-x2 and the other two are fixed,
so this is minus f, so σ Τ * so this is σ Τ 1 st you multiply and take
*.

(Refer Slide Time: 28:48)

On the other hand , what is Τ * f? It was, remember, here it is –f

but it is x2-, so it is f, but let me write what we got, okay, it is x2-
x3, x2-x1, x3-x1, right x3-x1, okay. So now let us apply, so we
have this, let us apply σ * to both sides, so σ * of Τ *f? What is
this see when apply σ *, I simply change subscripts by σ, so σ was
(13) right, so x2 –x3, 2 is fixed under σ, so it is x2 and 3 goes to 1,
so this is x2-x1, x2-x1, so 2->2 and 1->3, so this is x2-x3, 3->1
and 1 ->3, so you have x1-x3, and if you now compare this with σ
Τ*f okay.

So let us compare them both, we have x2-x1 , x2-x1, x2-x3,x2-x3,

so this term is common and this term is common and we have x1-
x3, and x1-x3. So we see that, so they are equal, they are same
right, so σ, so what we have is, σ Τ whole star f is applying σ * to

305
Τ * f. So you start with f, first apply Τ * then apply σ * to it. So
this is what we have in this example and I will now say this holds
in general, okay. So let me quickly say this. So in general we have
and why is this?

(Refer Slide Time 31:34)

What is the reason? And the calculations is the following. So what

is σ Τ*f? Which is by definition remember we take x and we apply
σ Τ of i –x σ Τ of j, remember again, when I for a permutation σ I
defined sigma * of f, how did we define this? Let us go back to the
definition of σ *, I start with f and each subscript I simply change
by σ, right.

So I change a subscript by σ , x σ I – x σ j. Here I am applying it

to σ Τ, so I change i by σ Τ, j by σ Τ. But now, this is same as
and the product is over this set i < j, i greater than equal to 1, j less
than or equal to n, but σ Τ remember that it is composition right, σ
Τ is the composition of σ and Τ, the product in, the product in the
symmetric group, the binary operation of the symmetric group is
the compositions of functions.

So this would be same as x times Τ of i x sub (σ of Τ of I) – x sub

(σ of Τ of j). I am not doing anything in this step, I am just
observing that σ Τ of i is σ of Τ of i, σ Τ of j is σ of Τ of j, but
what is this? This is applying σ to something, so this is σ * of
product 1<=i, i<j , j<= n of x Τ I – x Τ j, right, because if I take
this and apply σ *, applying σ * is by definition applying σ to each

306
subscript, so I am applying σ to each subscript, so this is σ * of
this.

But what is this? So this is σ * of this. But what is this, this is

exactly Τ * of product, the same product xi –x j right.

Because originally you had x i - x j, you apply Τ * to it and then

you get this. But this, what is this? This is just f, so this is σ * of Τ
* of f, remember this is exactly what I have claimed here, σ Τ * f
is σ* of Τ * f, okay. So, I do not need this interior parenthesis, so
just, okay, so in general I have this.

Online Editing and Post Production

Karthick

Ravichandaran

Mohanarangan

Sribalaji

Komathi

Vignesh

Mahesh kumar

Web Studio Team

Anitha

307
 Bharathi

Catherine

Clifford

Deepthi

Dhivya

Divya

Gayathri

Gokulsekar

Halid

Hemavathy

Jagadeeshwaran

Jayanthi

Kamala

Lakshimipriya

Libin

Madhu

Maria Neeta

Mohana

308
 Mohana Sundari

Muralikrishnan

Nivetha

Parkavi

Poonkuzhale

Poornika

Premkumar

Ragavi

Raja

Renuka

Saravanan

Sathya

Shirley

Sorna

Subash

Suriyaprakash

Vinothini

Executive Producer

309
 Kannan Krishnamurty

NPTEL Co-ordinates

Prof. Andrew Thangaraj

Prof. Prathap Haridoss

IIT Madras Production

Funded by

Department of Higher Education

Ministry of Human Resource Development

Government of India

HYPERLINK "http://www.mptel.ac.in"www.mptel.ac.in

Copyrights Reserved

310
 NPTEL

NPTEL ONLINE COURSE

Introduction to Abstract

Group Theory

Module 06

Lecture 31 – “Odd and even permutations II”

PROF.KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

Now I want to understand what σ star does when σ is a

transposition, so now we want to understand,
(Refer Slide Time: 00:28)

when σ is a permutation, a transposition. So let me remind you our

goal in this video was to prove the theorem that I stated some time
back, so maybe it is good time to recall the theorem.
(Refer Slide Time: 00:51)

(Refer Slide Time: 00:56)

The goal of theorem is, the goal of video, is prove this theorem. If
our permutation rho, and you have two different representations as

311
a product of transposition, they are both, σ’s are 2-cycles and tau’s
are 2-cycles then k and t are both even or both odd.
(Refer Slide Time: 01:14)

And in order to this, let me recap what we done so far, we have

defined a polynomial we have defined what is the modification of
that polynomial via an element of the symmetry group, σ*f and I
looked at various examples to explain what to σ*f, and then I
commented that when I take the product of two permutations and
apply * it is same as first applying first one,
(Refer Slide Time: 01:38)

and to apply the second one to the resulting polynomial, σ* then

apply to tau*f, okay, and I proved that. And hopefully it is clear to
you as we have seen in several examples.

Now I want to understand what is σ*f, when σ is a permutation.

Again we have seen this in our examples we got σ*f was – f, if
recall σ *f was –f, whenever we in our examples σ was a 2-cycle,
for example, in this example is σ is (13),
(Refer Slide Time: 02:25)

σ *f is –f, okay. Is this is always true? That if you have a

transposition,

(Refer Slide Time: 02:37)

312
then its * is – f? Yes, this is always true. I am going to prove this,
but let me first do an easy case of this first, and hopefully that you
will be convinced, then I will give you a more general proof.

So first of all take a special case, as a special case consider σ is

(1,2) and n is general here. So what is σ of F σ* of F? σ * of F
remember what it is. So F remember, first of all recall what is F, in
general now, f is (X1- X2), (X1-X3), …(X1 – Xn) then have (X2-
X3)… (X2 – X n), and then you have a other terms.
I am only interested terms which involves X1, other terms, right,
is this clear? We have X1 – X2, X1- X3, X1- Xn then X2- X3,
X2- Xn, then X3 – X4, X3- X5 and so on, but remaining terms do
not involve 1 and 2 right. So I do not care about that.
(Refer Slide Time: 04:17)

So what is σ * f? σ is (1,2). I am only interested to (1, 2). σ * of F

will be all, these first few terms will change now. So it will be X2
– X1, right then X2 – X3, up to X2 – X1, X2 – Xn, because the
first term 1 and 2 in are interchanged.

So it becomes X2 – X1, second term 1 becomes 2 so X1 becomes

X 2 and X3 is fixed so X2 – X3, X2 – X4, and then X1-Xn so that
X2 – Xn. What happens to the terms involves X2?

X2 becomes X1 right, so it is X1 – X3, X1 – X4, all the way up to

X1 – Xn, and other terms, other terms are fixed because other
terms are terms whose subscripts are different from 1 and 2. And
σ does not change them, so they remain the same, right. Other
terms remain the same.

313
So now what is the change now? X1 – X2 became X2 – X1 so
there is 1 minus sign, right, one interchange of the variables, order,
so X1- X2 became X2 - X1. But if you now look at this no other
interchanges happened. We just reordered them, X1–X3 comes
here, X1 – X4 comes here, X1 – Xn comes here, X2 – X3 comes
here, X2 – X4, X2 – Xn so all other terms, there is only one
change of sign,
(Refer Slide Time: 06:09)

right so this is – f. So remember I am claiming

(Refer Slide Time: 06:20)

that if σ is a transposition σ * f is – f, and I have proved it at least

in the special case that you have (1, 2) okay. So σ*f is – f.

Okay, now I am going to spend next the five minutes proving this
general and you can, hopefully you have understood what
happened in our examples and in this special case. This general
thing is important to prove, but if you are not, I mean it is more
important that you are convinced that is true based on these
examples, okay. Let me now prove in more general cases.

(Refer Slide Time: 07:00)

In general, we have σ * f = - f, if σ is a transposition, okay. This is

what I want to show. σ*f is – f, so I am going to assume, let σ, it

314
is a transposition right, in other words, it is a 2-cycle. So let us
say σ is (i j) with i less than j. I can always assume that because (i
j) is same as (j i), so i and j are different obviously, so I can put the
smaller one first and the bigger one in the second position. It is a
cycle so (i j) is same as (j i) so I can always arrange them so that
the first one is the smaller one.

And now let us investigate what is σ * f, okay. So let us

investigate what is σ * f? This involves, this requires us to
investigate what is σ * of a particular term in f okay, so consider.
(Refer Slide Time: 08:38)

a term of the form X u - X v right we have u < v remember these

are, because I used i and j denote σ, I have now used u and v to
denote a term, this is a typical term, right.
(Refer Slide Time: 09:04)

F remember consists of products X u – X v where the first index is

strictly smaller than the second index. In order to determine what
is σ * f, I need to see what it does to a term like this. Really what
we need to do is we have to figure out, to find σ * f, we have to
determine how many terms of the form change sign, right, under
σ, right, because if you know how many terms change the sign,
then we know what happens to F. If the number of changes is even
you have σ * f will be f, if the number of changes is odd σ * f will
be – f. So when does, the question is, so the question we want to
address is, the question is when does this change sign.

315
So when does X u – X v, change always under σ to X a – X b,
with b strictly less than a, right, when does this change sign when
you have a bigger thing minus a smaller thing. You start with a
smaller thing – bigger thing, so here of course use u is less than v.
You only start with such a term, u less than v, and see when it
changes to b less than a, the second term less than the the first
term. So in order to address this let us consider various cases.
(Refer Slide Time: 11:28)

If u and v do not belong to the set i, j, if u is different from i and j

and v is different from i and j, X u – X v does not change. That is
clear enough right, because if σ, what is, so remember σ * of X u –
X v is X σ u – X σ v. So we are now focused on when is σ u, for
what σ, σ u is strictly bigger then σ v, so we are interested in, for
which σu becomes strictly more than σv right, the first index
strictly bigger than second index.

The trivial case would be if u an v different from i and J then σ u

is u, σ v is v right, because if σ is (i j), if u and v both different
from i and j, σ of u is u σ v is v so this does not happen. Now let
us consider the other cases. So we have the remaining cases are
the following.

(Refer Slide Time: 13:15)

Case 1: so we can have u=i, and v=j, so remember u is less than v,
and i is strictly less than v, so you can u=i, v=j then what is σ * of
X u – X v, this is X σ u- Xσ v, but u is i, so this is X v- X u. So
this is one case were σ u is of course bigger than σ v, so in this
case the sign changes. So this is one situation where sign changes.

316
Let us take case 2, let take u to be i and j is something else okay,
so v is something else. So if you take u to be i.

So what we have is, let’s see, in this case, see remember I have
already dealt with the case that u is different from i and j, v is
different from i and j. So one of them must be u, and one of the
must be j. If one of them is i the other is j both happen then that is
case one. Other cases u=i, and v is not equal to j, let’s say, okay.

In this case what do we have? We have u=i, which is less than, we

have two cases, so let us suppose that, let us compare v and j.

So if v is less than j, so you have u=i right, so this is a sub-case of

case 2. In this case what happens to X u – X v under σ *, σ * of X
u – X v, this becomes, this is X of σ u – X of σ v. But u is i so this
is x of σ i which is j, and σ v, remember, v is different from i and j
so this is just X v right. In this case a change of sign happens.

15 minutes 56 seconds.

317
 Introduction to Abstract
Group Theory
Module 06
Lecture 32 – “Alternating groups”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

So this notion is well defined, because any two representations

have either even number of transpositions or odd number of
transpositions. So, now let us observe that even permutation have
an interesting property.
(Refer Slide Time: 00:30)

So, before that, let me quickly give you an example. So you have,
let us take S3, consider actually it is (12) (32). so this is in S3.
Right, you can immediately conclude that, σ is even. Because, it is
a product of 2 transpositions. Right, as we said in the theorem, it is
possible that it is also some other, maybe it is a product of some
other number of transpositions. But they must also have an even
number. So, σ is even. On the other hand, another example is, any
transposition is odd. Right, because if you take (12) it requires
exactly 1 transposition. So it must be odd.

(Refer Slide Time: 01:49)

What about a 3-cycle? I claim a 3-cycle is even. Why is this? So

example if you take (123) right, let us try to write this as a product
of 2-cycles. And we saw this yesterday, it is same as (12) (23). Is

318
this right? 2 goes to 3, so 2 goes to 3, 3 goes to 2 and 2 goes 1, so
3 goes to1, 1 goes to 2. So, this implies, 1 (123) is even. Of course
this is true of any 3-cycle. More generally, a K-cycle is even if K
is odd. And it is odd, if K is even.

So, this is a confusing situation. But it is clear right? We saw

earlier that a 2-cycle is dd. Because, 2-cycle is transposition. A 3-
cycle is even okay, 3 is odd but 3-cycle is even. Why is this?

Because if you take a K-cycle, and we have written this as a

product of transpositions earlier, and this turns out to be (i1 i2) (i2
i3) (ik-1 ik), right, and again let me repeat once more that if you
have one representation, as a product of transpositions, that will
allow you to determined whether the permutation is even or odd.
Because, every other permutation, must have a same number,
same parity, the number of permutation must have same parity.

Now how many transpositions are there? There are K-1

transpositions here. Right, because 1, 2, up to K-1. So this tells me
that if K is odd, implies K-1, so, this is σ, if K is odd, K-1 is even
implies σ is even. Because it requires K-1 transpositions. Of
course, if this is if and only, or you can write it like this, K is even
means K-1 is odd, so σ is odd. So, a K-cycle has, K-cycle is odd
precisely when K is even.
(Refer Slide Time: 04:38)

So, we can define now, sometimes it is useful to talk of sign of a

permutation. So sign of σ is either 1 if σ is even, -1 if σ is odd.
Okay, so this is the sign of σ. So, sign is just 1 if it even -1 if it is
odd.

319
So, let us come back to the even permutations and look at them a
bit more carefully. So, we want to prove a proposition, product of
two even permutations is even, inverse of an even permutation is
even.

So, if you take a couple of even permutations and take their

product, it is even. If you take an even permutation, and take its
inverse is also even. Why is this? If σ1 and σ2 are even, that
means σ1 can be written as a product of τ1 up to τr. Right, τk
where K is even. Similarly τ2 σ2 can be written as some ρ1 up to
ρt then t is even. What is σ1 σ2 now? This is τ1 τk, ρ1 to ρt. Right,
this is a product of transpositions.

So, if k and t are even, then k+t is even. Right, that is all we need.
If k and t are even k+t is even. This implies σ1 σ2 is even, because
the product decomposition of σ1 σ2 requires an even number of
permutations. This is even, this is even.
(Refer Slide Time: 06:56)

Now on the other hand if σ is even and say σ is σ1 up to σk. So, k

is even. What is σ inverse? I claim, σ inverse is σk inverse, σ(k-1)
inverse up to σ2 inverse σ1 inverse, right, this is clear, because in
general, in any group, recall in any group G, (ab) inverse is b
inverse a inverse. So, σ1 to σk whole inverse is, you take the
inverses and multiply in the reverse order. σk inverse … σ1
inverse. Now, if σ1 is 2-cycle, σ1 inverse is actually σ1. Because

320
remember (1 2) times (1 2) for example, is e. so this is σk. σk-1 …
σ2 σ1. So, these are all transpositions, σ inverse is also even.
Because, it requires again k many transpositions which is even.
So, σ1 inverse is even. Right, if you have a product of two even
permutation it is even. Inverse of an even permutation is even. So
because this is proof of the proposition.
(Refer Slide Time: 08:31)

So, now let us define a subset An. These are {σ in Sn, σ is even}.
So this the subset consisting of even permutations. Right, I am
taking all the even permutations. By the proposition okay, so
actually almost by the proposition, An is a subgroup of Sn. Let see
what is a subgroup of a group? It must include the identity
element. So I didn’t write it in the proposition. But I should say it
now. Certainly, what is the identity permutation? It is even right,
because it requires 0 transpositions.

So, this is 0 transpositions. We do not need any transpositions to

write it. So, identity certain even okay, right, if you think about it.
Identity is a certainly even. And if you take two elements of An.
The product is again in An. That was the proposition. If 2
permutation are even, the product is even, if you take something in
An, its inverse is also in An because it is even.
(Refer Slide Time: 10:08)

So, An is called the “alternating group”. And it is subgroup of Sn.

Okay, now let me just quickly mention one fact about An. Namely,
what is its order?

321
So, now I am going to introduce a homomorphism, consider the
group homomorphism φ from Sn to {1,-1}. So before I tell you
what φ is, just a word about what is {1, -1}. {1, -1} it is group
right, this is a group under multiplication. So, if you wish this is a
subgroup of nonzero rational numbers under multiplication, if you
wish. But abstractly, it is a cycle group of order to. Okay, so what
is the group homomorphism φ?

I will define φ(σ) to be sign (σ). Okay, if σ is odd, it goes to -1, if

σ is even, it goes to 1. So, first all of, φ is a group homomorphism.

Right, this is because, φ is a group homomorphism because of the

proposition. That we just did above. Right, because if φ(σ1 σ2), so
almost by the proposition actually, we need to be careful. So, what
is σ1 σ2. So this is sign σ1 σ2, right. Which is another way of
saying, you write σ1 as a product of transpositions. And see how
many there are. If it is odd sign is -1. If it is even sign is 1.

But by the same idea that we used in the proposition, we can first
write σ1 is a product of transpositions. And then write σ2 as a
product of transpositions. Put them together. If σ1 is I think I used
τ1 … τ k, σ2 is ρ1. … ρt, then σ1 σ2 is right, now if k is odd, t is
odd, k, t are both odd.

First let’s start with k, t are both even. Then k+t is even. k, t are
both odd. Then k+t is even right, two odd numbers add up to an
even number. One odd, one even right, exactly one of the 2
numbers k and t is odd. The other is even. Then k+t is odd. This is
the proof of this.

322
Because sign of σ1 is determined by whether k is odd or even.
Sign of σ2 is determined by whether t is odd or even. So in this
case you have sign σ1 and sign σ2 are both 1 and the product is
also 1. In this case we have sign of σ1 and sign of σ2 are both -1.
So, -1 times -1 is 1, which is the sign of σ1 σ2, because the σ1 σ2
is even. And here one is even, one is odd. So, it is 1 times minus 1,
which is -1.
(Refer Slide Time: 14:29)

Okay, so φ is a group homomorphism. I am going fast here may

be, but I hope this is clear enough. So, φ is a group
homomorphism. What are the other properties of φ.

So φ is onto. Right, because certainly both, because Sn contains

even permutations as well as odd permutations. Right, because it
contains both even and odd permutations. Clearly it is onto. Some
even things, for example, 2-cycles are odd 3-cycles are even. And
let us take, so here I am taking n at least. So I am going to let us
say n is at least, in all these considerations n is at least 2. If I take
n=1 it is not interesting. So always assume, because if you take
n=1, then there is only one permutation that is even.

So, now I claim that kernel of φ. What is kernel of φ? These are

the things that go to the identity element of {1, -1}. What is
identity element of {1, -1}? It is 1. So, all permutations which go
to 1. In other word, all permutation sthat are even. So, this the
precisely An. Right kernel of φ is An.

323
(Refer Slide Time: 16:11)

This is because An is precisely the set of even permutations. Now

the first isomorphism theorem says, Sn mod An is isomorphic to
{1, -1}. So, in particular, from the counting formula, says that |Sn|
is equal to |An| times cardinality of this group, which is 2. Okay,
and we know this is n factorial.
(Refer Slide Time: 16:59)

So, we have n factorial equals 2|An|. So this implies |An| is n

factorial by 2. So, upshot of this is, exactly half of the elements of
Sn are even the other half are odd. So, of the n factorial many
elements of Sn, half are even, half are odd. And even permutation
is are nice because they form a group. So, now it is just a quick
check for you. Can you also take odd permutations and see i they
form a group.
(Refer Slide Time: 17:57)

So, a quick question for you. Do odd permutations also form a

group? The answer to this is an oblivious no. For several reasons,
first identity is not there right, because identity is even. So identity
is not odd. This is one reason, that is good enough to conclude that
they are not a group.

But also remember if you take (12) and (34) are both odd. But the
product is even. So this amounts to saying that sum of two odd
numbers is even. So just like even numbers in integers form a

324
subgroup but not odd integers, similarly even permutations form a
subgroup, but odd permutation do not.

Okay, so this the end of my video about symmetric groups. So,

just for recap what we have done is, what we have learned how to
use cycle notation for permutations. We learned how to write a
given permutation as a product of disjoint cycles. Using that we
can figure out the order, the order of a K-cycle is K and order of a
product of disjoint K-cycles is the LCM of the orders of those
disjoint cycles.

Then we also saw that any permutation can be written as a

product of the transpositions, not necessarily disjoint and not
necessarily unique, but we proved a big theorem saying that the
number of transpositions required is always even or always odd.
Using this we have defined the notion of odd and even
permutations and we saw that even permutations form of subgroup
of the symmetric group. We called it the alternating group An. And
we finally saw that order of An is exactly half of the order of Sn.
So, An has order n factorial divided by 2. That means of the n
factorial permutations in Sn, exactly half are even, the other half
are the odd. Thank you.
Online Editing and Post Production
Karthik
Ravichandran
Mohanarangan
Sribalaji

325
 Komathi
Vignesh
Mahesh Kumar
Web Studio Team
Anitha
Bharathi
Clifford
Deepthi
Dhivya
Divya
Gayathri
Gokulsekar
Halid
Heamvathy
Jagadeeshwaran
Jayanthi
Kamala
Lakshmipriya
Libin
Madhu
Maria Neeta
Mohana
Mohana Sundari
Muralikrishnan
Nivetha

326
 Parkavi
Poonkuzhale
Poornika
Premkumar
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirley
Subash
Suriyaprakash
Vinothini
Executive producer
Kannan Krishnamurty
NPTEL Co-ordinators
Prof.Andrew Thangaraj
Prof. Parthap Haridoss
IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resource Development
Government of India
HYPERLINK "http://www.nptel.ac.in" www.nptel.ac.in
Copyright Reserved

327
328
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 06
Lecture 33- “Group actions”
PROF. KRISHANA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

In this video I am going to start with a very important notion about

group theory and this is the topic that will help us to prove the
remaining important theorems in this course, the final goal being
the Sylow theorems. So the important operation that I want to
define here, introduce to you in this video, is called group actions.

(Refer Slide time: 00:35)

Okay, so if you have not seen this before, this might seem abstract
and strange. So, I am going to define this carefully and do a few
examples, so that you are comfortable with this notion, it is very
simple notion.

So, let me first say that it is not difficult at all, it is a fairly easy
thing to understand, so group action, what is the situation? So, the
setup is the following.

329
So, we are going to fix a group which we will denote by G. Okay,
and we will fix set which we will denote by S. So G is a group and
S is a set. This set may depend will change according to the
context. It could even be the group G itself sometimes, the most
important examples that we will study S would be G itself, but the
important point to remember is S is just a set, even if it has some
other structure as far as the group action is concerned it is
irrelevant.

So, we want to understand the meaning of G acting on S. So, I am

going to define for you what is the meaning of G acting on S. So,
we say, I am going to write the definition which is very intuitive
and clear and explain it after writing it. We say G acts on S if there
is a function, okay, from G cross S to S. Okay, think of it like this.
What is G cross S? This is the Cartesian product. So recall G cross
S is simply all elements like this.

This is just a set, there are pairs the first one coming from G and
the second one coming from S. I denote the image by either gs or
g.s, or g*s, okay depending on the context, it is not important,
sometimes I will just write gs. May be I write g.s or g*s. But what
you have to remember is,
(Refer Slide Time: 03:32)

this function is, this function does what? It takes an element of G

and element of S, so this is the element of G I am calling g, this
element of S I am calling small s, so in this case I am going to
write small s okay, so I will try to keep this small s, to denote an
element of capital S, takes a element of small g of capital G, and

330
element s to capital S and outputs an element of capital S denoted
by gs, or g.s, or g*s, depending on the context.

So, this function is telling me, so this function is to be thought of

as, so G acts on S meaning G, capital G acts on capital S means,
you take a small element, element small s of S, small g of capital
G, and you tell how small g acts on small s and produce a second
element of the set S. But this function must have some properties,
it is not an action in general. This function has, this function must
have the following properties.

Really not has, should have, in order for a proper action we should
have the following properties, two properties. One is, if you apply
e, to small s you must get small s for all s in capital S, okay, e
being identity element, remember I am supposed to tell you what
is small g times small s, whatever it is e must map to s to s, and it
must be associative in the following sense: if you should take g
and g prime and apply to s that must be same as first applying g
prime to s and then applying g to s. So, this must be true for all
small s in capital S and small g and small g prime in capital G.

Okay, so we must have a notion of capital G acting on capital S,

but this notion must have some basic properties like the identity
element must act, must not do anything, so the identity element
must take an element to itself and how you apply for two distinct
elements of the group, whether you first multiply them and then
apply to s, or first apply one of the elements to s and then the other
element, you should get the same result. So this is the action, so
this is to be thought of as a very abstract situation, G is abstract to
capital S is abstract set meaning, it is not, it does not have any
properties, and S is just a set, and you have a function with these
properties.

331
(Refer Slide time: 07:31)

I am going to give you a some examples here, to illustrate this

action and to illustrate the point that, this should work in various
situations and different contexts this happens.

Okay consider, so first example. Consider an equilateral triangle.

Let’s take S to be vertices of it, and this is {A,B,C}. So the set is
{A,B,C}. So for every action we must have a set and a group and
the group I take to be the group of rotational symmetries.

So, if you recall it has two elements, three elements, R1 is

rotational by 120°, R2 is the rotation by 240°, okay. So, G is this. I
claim that, in the above sense, so whenever I say ta group acts on a
set, I always mean in the sense that I wrote in the beginning of the
video. Okay, so we must have a group and a set and a way to
attach, compose if you want to think like this, a way to combine
an element of the group and an element of set to produce another
element of the set, satisfying the two properties that I wrote here,
the identity element must fix every s and you must have
associativity.

For example what is R1? So what is e dot A? Identity means the

identity rotation. So, you do not change anything. So, this is A, E
dot B is B, E dot C is C right. So this is, the first property is
satisfied. What is R1 dot A? So R1 dot A, if you rotate, if you can
imagine by rotating by 120°, R1 of A is B, R1 of B becomes C, R1
of C is a A. Similarly, R2 of A is C, if you rotate by 240° A goes to
C, B goes to A and C goes to B, okay.
(Refer Slide Time: 10:59)

332
(Refer Slide Time: 10:59)

So this gives you a function from G, to S to S, so we get a

function satisfying the required two properties okay. So you can
check that, the second property is also satisfied. For example what
is R1 R2 of A? It is R1R2 of A. So first of all R1, R2 of A, can be
computed in two ways, so R1, R2 is remember rotation by 120 +
240 so that is identity. So this is A, if you first do R1 and R2 then
you get this, but now
let’s do R1 of R2 A, what is R2 of A?

In the function you see that R2 of A is C. So, this is R1 of C, and

this is R1 of C, R1 of C is A, so R1R2 of A is same as R1 of R2 of
A, so this as you can check with other examples, this will, the
associativity will hold. So just think for a second about this
example, so remember my goal right now is to explain to you
what is a group action on a set, the definition that I wrote it may
not be completely clear, so I want to give you three, four examples
at least to illustrate the definition.

So, the first example is this, and please pay close attention to this
and all the subsequent examples and hopefully after these
examples you will get a sense of what group actions are.

In this example we have rotations and we are only interested in the

set of vertices of this equilateral triangle, given any element of the

333
rotational symmetries, we apply it to the set of vertices and we get
another vertex, so this complete description of the action is given
here.

So, the group of rotations is acting on the set, so in some sense it

is just permuting the vertices. Okay, if you think about this, this is
related to the symmetric group because this is the way that
symmetric group also acts on these three things. So, this is a group
operation.

Let us look at one more example. Sn that we have studied in the

past which is a symmetric group acts, on almost by definition, it
acts on the set of indices, because how do you define σ? You take
σ of an element of Sn and you take i so this is my S. Okay, and G
is this. The set here is the set of indices and the group is the
symmetric group. So, I take a σ in Sn and I take i in S. Then what
is σ i, σ i is simply just σ of i, right.

(Refer Slide Time: 14:21)

So, for example, you take G to be S3; it acts on, take G to be S4,
it acts on S which is {1, 2, 3, 4}. If you take σ to be 1, 2, 3, let us
say (1, 4, 3, 2), this is an S4. What does it do to 1? It sends it to 4.
So, σ .1, this is a just a new notation but it is a same idea. The σ .2
is 1, σ.3 is 2, and σ.4 is 3.

334
Right, so σ, and remember the identity permutation fixes each i,
right, this is the first condition and certainly σ 1, σ 2 apply to i,
because σ 1σ 2 is a composition, is σ 1 applied to σ 2 of i. Okay,
so this is the associativity. So, both properties are satisfied.

So, you can say Sn acts on the set S. Okay, so this is symmetric
group acting on the set of indices from 1 to n is an important
example of group actions.
(Refer Slide Time: 15:57)

One more example. Let us take the group of invertible n by n

matrices. So these are n by n invertible real matrices. Right, this is
a group under multiplication that we have seen in the past and let
us take S to be R^n, so these are all column vectors of length n.
So, this is, for example, these are the things like so, this is in R^n
where ai’s are real numbers. How does this act? What is this
action here? So, now I want to say GL n(R ) acts on R^n in the
following way.

Okay, so you take a matrix and you take a element. Let us have v
in R^n. So, we need to tell what is A.v. But in this case there is a
natural operation right, so we can simply take A times, so this is
simply product of matrices. Right so this is an n by n matrix, this
is an n by 1 matrix. So, you can multiply
(Refer Slide Time: 17:53)

and what is the output? A.v is another vector or another element of

R^n. Which is how it should be, right? If you take a group G and a
set S, when you apply a group element to the set element you must

335
get another set element. So, I have taken a group element A here, I
have taken a set element v and I multiplied them to get another set
element. And one can check that the two properties of group
action are satisfied. What is the identity element? Of the group
GLn, it is the identity element, identity matrix.

And if you take identity times any vector any n by 1 column

vector, it is just v obviously right, because identity times any
matrix is itself. Similarly if you take AB times v, this is same as A
times Bv. This is again because matrix multiplication is
associative. So, we can say now that GL n(R ) acts on R^n. In all
these things I will come back and do more, look at these examples
again and again. At this point I am only giving you several
examples of group operations. Okay we have so far looked at three
examples. Now let us look at fourth example.
(Refer Slide Time: 19:28)

Let G be any group and take the set also to be G. So, we will see
how G acts on itself. So remember in general we talk about a
group G, we will talk about how a group G acts on a set S. A
group G acts on a set. In this example I am taking the group G
acting on G itself, but the second copy of the group is really a set.
When I am thinking of G as S the group operation of S is
irrelevant. Okay, there are two ways.

Okay, let me first say I will define the action as follows. So let us
take g in G and let us continue to use s in capital S but capital S is

336
also again capital G. So I define g.s to be gs, just the
multiplication.

So, this is the left multiplication by g. So, meaning you take g.s to
be s multiplied by g on the left side. There is also a right
multiplication that would be g.s to be defined as s times g, you
multiply by g on the right hand side but now let us look at left
multiplication.
(Refer Slide Time: 21:33)

So, the function from G to S, G cross G to G to G, small g, small s

goes to just the product. Okay, is this a group action? So this is a
group action or not? This is a function right that is okay.

Now let us check the properties of group action. Remember a

group action must satisfy this problem, e.s, what is e.s? Is by
definition just es, in the left multiplication, so this is s. So this is
okay. If you take g1g2.s, this is also okay because the
multiplication in the group G is associative. So this is a group
action. So in all these examples, please note that there is
something I am doing. It is not just a formality.

In each example the property that I have to check is true because

of some existing properties that you have learned. In the first
example, because rotations are compositions, multiplication of
rotations is a composition and composition of functions is
associative, we can say this. It is not just a formality; it is not blind
symbols I am writing here.

337
In the second example also product of permutation is by
definition, composition of permutations which are functions. So
composition is associative. So again it happens to be true, the
associativity.

In the third example associatively holds because matrix

multiplication is associative, that is an established fact from
before. Okay, you have learned it in some other course or it is an
established fact in some other course. So because of this existing
known fact, these are true. Similarly here group multiplication is
an associative operation by definition of a group.

So this is an existing fact which tells us that this is a group action.

This action is called left multiplication of G on itself. Clearly
enough right. This is an action of G on itself by left multiplication.
This is an important action that we will see again in the course.

Online Editing and Post Production

Karthik
Ravichandran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh Kumar
Web Studio Team
Anitha

338
 Bharathi
Clifford
Deepthi
Dhivya
Divya
Gayathri
Gokulsekar
Halid
Heamvathy
Jagadeeshwaran
Jayanthi
Kamala
Lakshmipriya
Libin
Madhu
Maria Neeta
Mohana
Mohana Sundari
Muralikrishnan
Nivetha
Parkavi
Poonkuzhale
Poornika
Premkumar
Ragavi

339
 Raja
Renuka
Saravanan
Sathya
Shirley
Subash
Suriyaprakash
Vinothini
Executive producer
Kannan Krishnamurty
NPTEL Co-ordinators
Prof.Andrew Thangaraj
Prof. Parthap Haridoss
IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resource Development
Government of India
HYPERLINK "http://www.nptel.ac.in" www.nptel.ac.in
Copyright Reserved

340
341
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 06
Lecture 34- “Examples of Group actions”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE
(Refer Slide Time: 00:17)

Let us do one more example, so again this is another example of G

acting on itself. This is also an extremely important example of a
group acting on itself. So what is this? So this is called
conjugation, action by conjugation. So, given g in G and s in G
defined g*s, I am going to use star here to be very clear about the
action, this is star refers to the action. Not the product insides the
group.
(Refer Slide Time: 01:02)

Earlier I could be slightly vague because action is already the

product inside the group.
(Refer Slide Time: 01:07)

Here, I defined it to be g*s is gsg inverse. So gsg inverse is called

conjugation of s by g. So this is called the conjugation of s by g,
this is the conjugate of s by g. So I should not write conjugation, it

342
is called conjugate. The process of doing this is called
conjugation. So is this is an operation? Is this is an action? Let us
check this, is this an action? So you have e dot or, e*s, it is by
definition, so the action is defined to be this: e s e inverse which is
ese which is s. So that is okay. What about g1 g2 *s by definition
this is same as g1g2 times s times g1g2 inverse. Right this is the
action.

(Refer Slide Time: 02:27)

But this is same as (g1g2) s g2 inverse g1 inverse right, because

that is inverse in a group, product of g1g2 inverse is g2 inverse
times g1 inverse. This is g1 g2 s g2 inverse, by associativity of
group operation. And this is g1 (g2*s) g1 inverse, right, because
g2*s is by definition g2sg2 inverse. But this is same as g1*(g2*s),
because g1*something is g1 times that thing times g1 inverse,
right, this is the operation. So g1* this middle thing, g1 times the
middle thing times g1 inverse, that g1* that middle thing.

Right so this says that g1g2*s is g1*g2s. So * is an operation, is

an action. Again what I am doing is not a formality okay; it is,
there is some content in this. * is an action, the identity is okay,
that is very clear but the associativity required a small proof, right.
(Refer Slide Time: 03:59)

So this is called, this action of G on itself, is called the conjugation

action. Okay, so the two actions that we have studied by G on

343
itself, or the left multiplication and conjugation and both are
extremely important okay, both are important because of the
results that we can prove using these actions later okay, so this is
the, for now let me stop with the examples here. There are some
other examples that I will come back to later.

But now let us see some properties of group action. So now let us
continue with our study of, as I said this is actually an easy notion,
group actions, initially if you are seeing it for the first time it is
somewhat strange. So it will take some time get used to, that is
only difficulty the unfamiliarity is the difficulty.
(Refer Slide Time: 05:33)

So please understand the examples very carefully, so I have given

five examples. First one being, how rotations operate on the
vertices of equilateral triangle. Then our familiar symmetric group
acting on in indices. Invertible matrices acting on column vectors.
And finally the 4th and 5th examples were a group acting on itself
first by left multiplication and then by conjugation.

So these are the important examples. And if you understand all the
examples carefully, it is also illustrates the point I made earlier
that group action supposed to be abstract, it is not in some specific
context. In any context we have a group and a set S, and a
function from G cross S to S satisfying some conditions we have a
group action. So now there are some obvious, some important
things that we can attach to group actions and they are the
following.

344
(Refer Slide Time: 06:21)

So let G, so the general situation is, let G be a group acting on a

set S. So G is an abstract arbitrary group. S is an abstract arbitrary
set. Let s be in S, the orbit of s, denoted by os, orbit o for orbit, is
all, so you take gs okay, see the word orbit, you know the word
orbit, it is how an object moves around right, the orbit of earth
around the sun for example. It is how earth moves around the sun,
you connect all the points when it is travels and you get the orbit.
The same idea works here.

You take a small s, orbit is always defined for an element of the

set S, and see where it travels under the action of G. So you take
one group element and see what is gs. So I am going to use gs, to
denote the action. So, here we have G cross S to S. I will take a
small g, small s and I will denote its image by gs. I will not use *
or (.) just for simplicity. So, I take small s see wherever it travels
under the action of capital G. So, small g, one element small g,
take another element g prime take g double prime. You do this for
every element small g and see where it travels and you see call
that orbit of s. Orbit of s is of course a subset of capital S.
(Refer Slide Time: 08:29)

So, orbit of s should be thought of as the set where small s travels

under the action of G, okay. Now I should quickly check, in all the
examples orbits. For example, okay, so the first example G was
the group of rotations and S was the vertices. What is the orbit of
s, of let us say of A? Now if you go back and see the orbit, action,

345
A will always be in the orbit because, identity A will be A. And r1
A was B, r2 A was C.

So orbit of A in this case happens to be S. Similarly this is also

orbit of B and orbit of C. In the second one G was the symmetric
group, and S was indices 1 through N. What is the orbit of 1? 1 is
there certainly because the identity 1 is 1, 2 is there because (1,2)
sends 1 to 2, 3 is also there because (1, 2) applied to 1 is 2, (1,3)
applied to 1 is 3. (1 i) in general applied to 1 is i. So in this case
also orbit of 1 is all of S, and S you can see this is orbit of 2, orbit
of 3, orbit of n. So they are all S.
(Refer Slide Time: 10:42)

In the third example we have G=GLn(R), and S=R^n. If you take

the zero vector, okay, in the previous two examples you might
think that orbit is always equal to all of the set. But that is not
necessarily the case. So what is orbit of, if you take any matrix, A
and multiply by 0, you get 0 right, so orbit of 0 is just the element
0, nothing else is there. And now on the other end to take a vector,
this is an exercise that you can do using properties of matrix
multiplication. If V is R^n and V is different from 0 then the orbit
of V is all of R^n minus 0.
(Refer Slide Time: 11:53)

So this is to say that, in other words, given any nonzero vectors, v

and w in, so I do not have a bar here, w in R^n there exists an
invertible nXn matrix A such that A.v=w. So what I am saying is
that, let us take this as a fact okay, this fact can be independently
proved. It is not part of what I am now taking about so I do not
want to spend time on this. But what does this fact mean for the

346
orbit of a vector v is what I am interested in. Suppose this
statement is true, given any to nonzero vectors v and w there exits
an invertible matrix A such that A.v=w.
(Refer Slide Time: 13:08)

Then this implies that orbit of v, remember in order to get this w

has to be nonzero also is all nonzero vectors. In R^n in other
words it R^n minus the element 0. Okay, because if you recall
what is orbit of v, by definition it is A.v as A varies inside the
group, in this case GLn. So you take A.v as A varies but because
of this fact you give me any w in R^n minus 0, there is an A such
that A.v=w. So every w will be in the orbit. So in this case there
are two orbits: 0 vector is an orbit by itself, and all nonzero
vectors are an orbit by itself, one orbit together.
(Refer Slide Time: 14:05)

So, now G acts on G by left multiplication right. In this case what

happens? So take an element s in G. What is the orbit of s? Let us
calculate this, orbit of s is gs, as, because gs is just in this case
action by left multiplication, so when I write gs I mean actually
the multiplication in the group.

It is gs has g varies over G. We claim that orbit of s, so again it is

s, okay, is all of G. Why? So the proof of this. So a priori this is a
subject of G right, it is a subset of S which is of course G, orbit is
a subset of S. In our example, it is G. Why is orbit of s equal to G?
So take any t in G okay, let t be any element of G.

347
Can we find a group element g such that gs=t? See again let me
remind you, orbit is the subset of capital S through which small s
travels under the action of G. So s fixed, whenever we talk about
orbit the small s which is an element of S is fixed. And you are
applying all group elements to it. And what you get is the orbit. So
now in order to prove that Os=G, take any element of G which I
am calling t and we want to find a group element g such that gs=t.
(Refer Slide Time: 16; 22)

Of course we can find such an element. What is g? g is simply ts

inverse right. If you take that, because here G is playing two
avatars here remember. G is the group as well as the set. If you ts
inverse time s. What do you get? This is ts inverse s which t. So
every element, so this means that every t is in the orbit. So Os=G.
So again the orbit is all of G.

And finally let us look at the last example, G acts on itself by

conjugation. And this is more interesting okay, here the action
remember, is g*s, I am going to now use * because here it is not
multiplication of group, gsg inverse. Right, so recall. Now let us
look at the orbit, for example orbit of e, Oe, is I take arbitrary
elements of g and apply *e. This is by definition of the action g e g
inverse, as g is in G. But geg inverse is e for all g, so geg inverse
is e for all g in G. So this is just e. So orbit of e is e.

Now let us go back to the pervious example, I said orbit of every

element is equal to G. And in fact I should probably spend a
minute more on this. Just to illustrate this. What is orbit of e in this
example? It is ge has g varies in G. This is just G, okay, so I have
proved that the orbit of every element is G, so this checks out

348
here, the orbit of e is G, in particular orbit of e is G. But whereas
conjugation behaves very differently. Orbit of e is just e.
(Refer Slide Time: 18:57)

Okay, now suppose for a moment that G is abelian. Then what is

the orbit of, let us take an arbitrary element s in G. Then Os is by
definition g*s, g in G. This is by definition gsg inverse g in G. I
have assumed that G is abelian. If G is abelian this is same as sgg
inverse g in G. But then sgg inverse is s right, this is true for all g
in G. So this is just s, so orbit of s is s. So if the group G is
abelian, orbit of an element is just that element. So these are the
five examples, in the fifth example we get something interesting if
the group is abelian.
(Refer Slide Time: 20:14)

So now coming back to orbits, hopefully it is clear by these

examples, if G is a group an arbitrary group acting on a set S, if
you take a s in S, small s always contain s right. So orbits are
never empty and they always contain the elements.

Why is this? Because remember Os is gs as g varies, so in

particular, can take g to be e. So es which is s belongs to Os
always okay, orbits contain small s but they could be just small s
or they could be all S. So in this example of abelian group and
action by conjugation the orbit of an element is just that element,
nothing else, whereas in the action by left multiplication orbit is
all of the group, all of the set.

349
In the action GLn(R) on R^n, orbit of the zero vector is just the
zero vector, but orbit of any other nonzero vector is the set of all
nonzero vectors. And in these two actions the action of rotations
on the three vertices or the action of Sn on the indices 1 to n,
orbits are the entire sets.

So orbits are at least going to contain small s, orbit of small s

contains at least small s and it can vary in size depending on the
action.

So we have seen examples of different kinds of orbits. So in the

next videos we will study further about orbits, and other important
properties of actions and using these properties we will prove our
main theorems in the rest of the course. Thank you.
Online Editing and Post Production
Karthik
Ravichandran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh Kumar
Web Studio Team
Anitha
Bharathi
Clifford
Deepthi

350
 Dhivya
Divya
Gayathri
Gokulsekar
Halid
Heamvathy
Jagadeeshwaran
Jayanthi
Kamala
Lakshmipriya
Libin
Madhu
Maria Neeta
Mohana
Mohana Sundari
Muralikrishnan
Nivetha
Parkavi
Poonkuzhale
Poornika
Premkumar
Ragavi
Raja
Renuka
Saravanan

351
 Sathya
Shirley
Subash
Suriyaprakash
Vinothini
Executive producer
Kannan Krishnamurty
NPTEL Co-ordinators
Prof.Andrew Thangaraj
Prof. Parthap Haridoss
IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resource Development
Government of India
HYPERLINK "http://www.nptel.ac.in" www.nptel.ac.in
Copyright Reserved

352
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 07
Lecture 35-“Orbits and stabilizers”
PROF.KRISHNA HANUMANTHU
CHENNAI MATHAMATICAL INSTITUTE

Okay, let’s continue our study of group actions, in the last

videos I have defined what it means for a group G to act on a set
S. I have told you what orbit of an element is, orbit of an
element is the, so you take an element of capital S, its orbit is,
you apply all elements of the group to that fixed element, small
s, and you consider the set.

Consider the set of where it travels, so orbit of s is collection of

gs, where g varies over the group G, small g varies over the
group G. And we saw several examples of group actions and in
some of those examples we have seen orbits of specific
elements. Okay, so today I am going to start with thinking of
orbits as equivalence classes in the following way.
(Refer Slide Time: 01:15)

Okay, so let us say our set-up is the following. So G is a group,

S is a set, G acts on S right. This is our set-up always, a group
acting on a set. Define an equivalence relation, first let us say a
relation on S as follows: so I am going to define a relation
denoted by this symbol as follows. I will say that s1 is related to

353
s2, so we say s1 is related to s2, so remember we read this as “s1
is related to s2”, if there exists an element in the group G small
g, such that gs1 is s2. Remember gs1 denotes the action of small
g on small s1, so that is how we denote the action. So remember
the action means there is a function from G cross S to S, which
takes an element (g,s) to gs.
We just denote it simply by gs, instead of writing g star s, or g
dot s, we, for convenience, denote the action by gs. So now this
is the relation I am defining, s1 is related to s2, if they exists a
group element such that gs1 is equal to gs2.
(Refer Slide Time: 03:27)

So now the question is, is this an equivalence

reaction?Remember an equivalence relation is a relation
satisfying some conditions, right, what are they? We want every
element to be related to itself, you want the relation to be
reflexive.

So the first condition: is s related to s? Yes, because, why is s

related to s? You can take es to be s, you have es is equal to s,
this is by the definition of action, so you see that we need this, in
order to make this an equivalence relation, we must have this.
And this is guaranteed by the definition of group action itself. So
identity element sends s to s, so s is related to s. In order for s1
to be related to s2, we want a group element such that gs1 is
equal to gs2. We take the identity element of group which
always exists in a group and es is equal to s, so s is related to s.

Suppose if s1 is related to s2, then is it true that s2 is related to

354
s1? This is the symmetry property right, if s1 is related to s2, is
s2 related to s1, so let us see. If s1 related to s2, by definition,
there exists g in G, such that by definition this happens, right?
If s1 is related to s2 there exists a group element such that gs1 is
equal to gs2. Now let’s apply the action properties, and let’s in
particular apply g inverse to both sides, right? If gs1 is equal to
s2 remember these are elements of capital S, these are two
elements of capital S that are equal to each other. Let’s apply g
inverse to both of them, then by definition of group action, this
follows from as before definition of from group actions. What is
the definition? Remember group actions must satisfy two
condition, es is equal to s, and g1g2 of s is g1 of g2s.

These are properties of a group action these are the defining

properties of a group action. So g inverse gs1 is same as g
inverse g s1 but that is es1, g inverse g is e, but es1 again by the
definition of action is s1, so s1 is equal to g inverse s2. So that
means there is a group element such that that element times s2 is
s1, so s2 is related to s1. Right, remember we say that s1 is
related to s2, the first one is related to the second one, if a group
element takes the first one to the second one, here s2 is taken by
g inverse to s1, so s2 is related to s1.
So the question has an affirmative answer. If s1 is related to s2
then s2 is also related to s1.
(Refer Slide Time: 07:00)

This is the symmetry. Finally we want to check transitivity of

the group action. So suppose s1 is related to s2 and s2 is related
to s3, so that means g1s1 is s2, g2s2 is s3, for suitable g1,g2 in

355
capital G. Because s1 is related to s2, g1s1 is s2 for some group
element, s2 is related to s3, so some group element takes s2 to
s3.

Now again we apply g1, g2 to s1, by the definition of a group

action, this is g1 of g2s1, but g2s1. Let’s actually apply g2g1, g1
s1 by hypothesis s2, and g2 s2 by hypothesis again s3. So this
implies s1 goes to s3 with a group element g2g1. So s1 is related
to s3.
(Refer Slide Time: 08:39)

So we conclude, we checked the three defining properties of an

equivalence relation, so this is an equivalence relation. Right, so
again let’s recall what is the equivalence relation? It says that
two elements are equivalent, if there is a group element that
takes one to the other. So that is an equivalent relation I have
checked so far.

But now I am going to use, the definition, the important property

of equivalence relation, recall that, an equivalence relation in
some sense the most important property for an equivalence
relation is that it partitions the set S into disjoint equivalence
classes. We take equivalence classes.

When I did equivalence relations and when we talked about

cosets in a group, I did this in the detail, but so let us not repeat
it again. But an equivalence relation partitions a set into disjoint
equivalence classes, you take some element and you look at all

356
elements that are equivalent to it. That is an equivalence class.
And the most important property of equivalence class is that,
two equivalence classes are either disjoint or they are identical
to each other, right. They cannot be different and have
something in common.

So, if you take disjoint distinct equivalence classes they are all
disjoint and because every element is, its on equivalence, is in
its own the equivalence class, equivalence classes certainly
cover all of S. Okay, so now in this example of an equivalence
relation, now that we have a group action on a set, let us take a
small s in capital S, what is the equivalence class? Let us
compute the equivalence class of s.

Okay, in the earlier situation we have denoted equivalence

classes by this symbol, by definition what is this? These are all t
in S such that s is related to t. All elements of the set such that s
is related to t. But now I am working with a specific equivalence
relation.

So, these are all t in S such that t is gs for some g in G. Right,

remember this is the equivalence relation. We say that s1 is
related to s2 if there is a g such that gs1 is equal to s2. So, if t is
supposed to be related to s then gs must be t for some g. So, this
is simply can be written as gs for all g in G, right. I can
eliminate the variable t here or as small g varies in capital G, I
will take small gs. But we have another name for that.
(Refer Slide Time: 11:46)

357
From the last video, but this is simply the orbit of s right, so this
is simply the orbit of s. So the equivalence class of an element is
precisely, in the notation of the last week, of the last video, is the
orbit of s, right, so we have orbit of s remember is exactly this
set.

It is where s travels under the action of G, so equivalence classes

for the relation that we have defined today, is simply the orbits.
So we conclude hence, after all this, we have, we conclude that,
so the set-up is if G is a group acting on a set S, then the set S is
partitioned, you take the distinct orbits under the G-action on S.
This is an elaborate sentence and it is the conclusion of what we
have done so far.

If G is a group acting on a set S, then S is partitioned into

distinct orbits by the action G-action S. So in other words, S is a
union of orbits, so that we know, because every element is in its
own orbit, and two distinct orbits are disjoint. This is because
being related under G action is an equivalence relation so
distinct orbits are disjoint.

So we have the picture as I have drawn earlier, so you have a set

S, so you have Os1 that’s one orbit, Os2 in general an
equivalence relation partitions a set. In our situation equivalence
classes are orbits, okay, and so on.

This could be infinite of course, I am trying to just illustrate this,

orbits, number of orbits may be infinite, but each individual
piece in this partition is a single orbit, and they have nothing in
the common, two distinct orbits have nothing in common.

358
This is the first important observation about group actions that
we will have exploit a lot, so this is an important fact. So this is
an important observation as we will see later, we will make use
of this observation in the future videos. Okay, so now let me
introduce another important notion, object attached to a group
action, so let, now consider a group; I will come back to this
partition of a set into its orbits later.
(Refer Slide Time: 15:24)

But now for the moment, consider a group G acting on a set S.

Again let me repeat you, repeat for you, G is a group acting on a
set S, so G is an abstract group, it can be any group, and S is just
a set. Even when S is equal to its group itself, it is really is just a
set.

So for, let us fix a small s in capital S, I have defined for you the
orbit of s, Os, now I will define something else, the stabilizer of
s is the set of elements, is the subset of G, defined as, okay, so
stabilizer is denoted by G sub s, remember we have orbit of s
was denoted by Os.

Now I am defining stabilizer as G sub s or sometimes I will use

stab of s just for clarity, this is a subset of G, unlike the orbit,
orbit is a subset of the capital S. Stabilizer as in the definition
it’s a subset of, it is more than a subset actually, but we first
define it as a subset. It is all group elements that stabilize okay,
think of this as stabilizing s. So g does nothing to us, g fixes s,
so this is the set of group element that fix s, okay.

359
So this is group elements that fix s. So the immediate Lemma
that I will prove immediately Gs is not merely a subset of the
group, it is a subgroup of G.

What is the proof? We have now done so many examples and

results to check a group a subset is a subgroup, so let us quickly
do this, this is very easy. We have to check that the identity is
there and it is closed under multiplication and it is closed under
taking inverses.

First property: e times s is s, by the definition of group action.

So e fixes s, so e is in the stabilizer, stabilizer is all g such that
gs are equal to s. So stabilizer contains the identity element.
Let’s say g1& g2 are in the stabilizer of s, that means g1s is s,
and g2s is s, right. This is the definition of being in the
stabilizer, now what is g1 g2 of s, this is, by the definition of
group action, is equal to g1, g2 of s, which is equal to g1 of s,
because g2 of s is s, g1 of s is also s.

So this implies that g1g2 is in Gs, right. If two elements are in

the stabilizer, their product is in the stabilizer, that’s good. So
the stabilizer is closed under multiplication.
(Refer Slide Time: 19:11)

Let’s say, you take an element g in stabilizer, then what is gs?

That is s, by the definition of being in the stabilizer. Now if you
multiply both sides by g inverse, I get this, but this is same as g
inverse g of s is g inverse s, but that is es, is equal to g inverse s,

360
this means s is equal to g inverse s, right. But this means that g
inverse is in the stabilizer of s, because g inverse also fixes s, g
inverse s is s, so g inverse is in the stabilizer. So this is the proof
that stabilizer is a subgroup of G.
(Refer Slide Time: 20:05)

So I want to make this important note here. So given an action

of G on S and an element small s in capital S, we have two
things, we have orbit of s, which we denote by Os which is all gs
as g varies over G, this is inside S obviously. And we have
stabilizer of s that I sometimes denote by G sub s, sometimes by
Stab s, this is all g in G such that gs is equal to s. This is a
subgroup of G, right. This is the situation now.

So we have attached to a single element of capital S something

called the orbit which is a subset of capital S, remember capital
S is just a set, so we cannot talk about a subgroup or anything
like that, it is just a subset. Stabilizer is a subgroup of capital G.
So there are two things, so stabilizer and I will do, in a later
video, examples of stabilizers in more details.

361
 NPTEL
NPTEL ONLINE COURSE
INTRODUCTION TO ABSTRACT
GROUP THEORY
Module 07
Lecture 36 – “Counting Formula”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

Now let us now come to how these two are related.

(Refer Slide Time: 00:28)

This is an important theorem. So let us say G is a group, always

this is our set-up in these videos, G is a group and S is a set, G
acts on S, okay. So, then in this situation there exist a bijective
function φ: G mod the stabilizer to, okay, fix, let a small element
s be taken in capital S, okay let me repeat the theorem. G is an
arbitrary group acting on an arbitrary set S and let us fix an
element of the set, small s, so s is an element of S, then there
exists a bijective function between orbit of s and G mod Gs, and
what is G mod Gs, G is a group, Gs is a subgroup, so G mod Gs
is simply the set of left cosets of Gs in G. I must remark here
that this is not a group because Gs is not necessarily a normal
subgroup okay, so this is just a warning to you that, when I write
G mod Gs, typically used only when you have a normal
subgroup, and you consider the quotient group, but here I am
using the notation for convenience, to denote the left cosets of
this subgroup in this group, so in this theorem, you only think of
G mod Gs as a set of left cosets. So the proof is not difficult at
all.

362
Okay, with all the set-up that we have so far, so G mod Gs, there
is a bijective map from G mod Gs to Os, so, and actually define,
this should also be part of the theorem, define φ of, what is a left
coset, it is of the form gGs, G mod Gs is the set of left cosets, so
it is g times Gs, define this to be gs, this is of course in the orbit.
So what is the function? It is taking a left coset which by
definition is o f the form gGs and I will map it to the element gs
which is in the orbit.

So we have to check first of all that it is a well defined map,

because a left coset can have several representations with a
different g here, okay, so φ first we must check is well defined.
φ is well defined, so why is that? Suppose gGs = hGs, we must
ensure that φ(g) = φ(h), that is, we need to check gs = hs, this
obviously needs to be checked, because we are trying to define
that image of gGs is gs, but it can very well happen that
gGs=hGs, but g and h are different because several cosets will
coincide right.

But then if gGs is equal to hGs, but gs is different from hs, then
there is a problem, the function is not well defined, so we need
to check this.
(Refer Slide Time: 5:52)

But this is not difficult at all, if you start working this out, it will
be clear, so if gGs is equal to hGs then what do we have, this

363
implies that g = hg prime, for some g prime in Gs, right this is
clear to you right, if you have left cosets are equal, then I am
going to quickly work this out, so in general if you have G is a
group H a subgroup, if you have g1(H) equals g2(H), remember
g1 belongs to g1(H) because g1(H) is the set of g1 times every
element of capital H.

So, in particular g1 times identity will be one of the elements of

g1(H), which I am assuming is equal to g2(H) so g1 belongs to
g2(H), so g1= g2 times h for some h in H.

I am using the same idea here, so gGs=hGs, so g must be

something h times some in Gs, right. Now this implies, if I do
gs, now let’s apply, bring in the set here, the element s here, gs
equals h(g prime)s, because g is equal to h(g prime), but the
group action property says that h(g prime)s is equal to h(g
prime s), but g prime remember, where is g prime? g prime is in
the stabilizer.

Since g prime is in the stabilizer, what is the stabilizer means,

that is all elements in G that fix s, so g prime is an element of
the stabilizer means g prime fixes s, so g prime s = s, so h(g
prime s) is just hs. So we have concluded that g s equal to h s as
required, so if two left cosets gs and hs as required, so there will
be no problem you can choose one representative and I can
choose another representative, and we apply both the function φ,
we both get the same answer. So gs = hs, so that is good. So φ is
well-defined.

(Refer Slide Time: 8:09)

364
Now we are trying to say that φ is a bijection, so φ is well-
defined is all well and good, but we want to check that it is 1-1
and onto. Why is it 1-1? φ is 1-1, let’s check that. What does it
mean?

1-1 means if φ(g Gs) = φ(h Gs), suppose this happens we want
to conclude that g Gs = h Gs. Let us work it out, φ(g Gs) = φ(h
Gs), I am taking two elements to the left hand side, domain,
which is G mod Gs, so I am taking two elements g Gs and h Gs,
suppose, φ of these two elements are equal then that means gs =
hs because φ(gGs) = gs, but this means if you now say, h inverse
g is this, okay, this is something you have done before, we
multiply by h inverse on both sides on the left hand side, so h
inverse gs is equal to hs, h inverse h s which is s. h inverse g
applied to s is s.

Two three steps are missing here, I am not writing all the steps
because we have done these things before, but this means h
inverse g fixes s, so it is in the stabilizer of s, so h inverse g is on
the stabilizer of s, but h inverse g is in Gs, so is (h inverse g)
times Gs equals Gs, remember that if you have an element in the
group, the left coset determined by that element is the identity
left coset, so (h inverse g)Gs is equal to Gs, but this means
gGs=hGs, right.

So we have a function from G mod Gs to orbit of s, if two things

map to the same thing, we are saying that they are equal, so this
is 1-1, if g(Gs) and h(Gs) map to the same element, namely φ(g

365
Gs) = φ(h Gs), we concluded that gGs=hGs, so φ is 1-1, this
completes that proof.

Finally, is φ onto? What is an element in orbit of s? It is of the

form, a typical element, it is simply of the form gs, then we
have, you take the same g, φ(gGs) = gs, so onto-ness is very
clear here, because elements of the orbit of s are by definition of
the form gs, now for the same g, g(Gs) maps the gs, so φ is onto
and this completes the proof that φ is a bijective map.

Right, so this is how we relate the stabilizer and the orbit, there
is a set-theoretic bijection, let me emphasise that, because Os is
just a set, G mod Gs in general just a set, so there is just a
bijective map from G mod Gs to Os, namely given by a coset
gGs maps to gs, right this is the important definition of the
function, and once you have the function defined like that it is is
relatively easy to check that is is well-defined, it is 1-1 and it is
onto.

So now let us see, what are the applications of this theorem. So

the theorem says that, I called it a theorem I guess, it implies the
cardinality, so now I am going to use this symbol, it denotes the
number of elements, and okay we have used that to denote the
order of a group, let’s just extend it to denote the order of any
set, the cardinality of any set.
(Refer Slide Time 13:00)

366
So the number of elements of a set is denoted by vertical bars, so
if two sets are in bijection then the number of elements in those
two sets are equal. This will be most useful when the group and
the set are finite so, let us assume that, G is finite. So we need
only G is finite, orbit will be fine, S may not be finite, it doesn’t
matter. So I am going to assume G is finite.

Recall the counting formula, what was the counting formula that
we proved when we talked about cosets of a subgroup in a
group, if G is group, H is a subgroup, we said that the number of
left cosets is equal to the order of the group divided by the order
of the subgroup. So G/Gs, which was denoted in our earlier
videos by this, the index of, this is the index of Gs in G, this is
equal to order of G divided by order of H, right. So let us,
written differently, we have order of H, sorry H is Gs here, so
Gs times the index of Gs in G is order of G, right. So I am just
rewriting the counting formula.
(Refer Slide Time 15:22)

In our case, in our situation the index of Gs in G, is by the

theorem that we just proved, remember this is all relative to all
of this is relative to the fixed element s in capital S. So every
time you see small s here I am referring to that fixed small s, so
in our situation the index of Gs in G is the orbit size.

So we have the order of the stabilizer times the size of the orbit
is equal to the order of the group, okay. This is a very important
application of the theorem that we proved and the counting
formula.

367
This is also called a counting formula, this is in fact very similar
to the counting formula that we had earlier, namely G the index
of a subgroup in a group G is the ratio of the orders of the two,
subgroup, group divided by subgroup, applied to this particular
of the case of the group action on a set we have cardinality of
the stabilizer times cardinality of the order is equal to the
cardinality of the group. This is very important and has many
applications, we will see these in later videos, let me only say at
this point that, for example, we can immediately conclude that
the size of an orbit, let me say that the number elements in an
orbit must divide the order of the group.

This is not really clear, right, in general. Why would, so if a

group G acts on a set S, you took an orbit which is a subset of
this set S, why should the size of that orbit be related to order of
the group? It is not clear, right. Group is here acting on a set S,
orbit is inside the set, why should the size of the orbit be related
to the order of the group? And this is a consequence of the
counting formula. It is not only related, it is related in a very
precise way: size of the orbit must divide the cardinality of the
group, the size of the group.

So this is a very useful observation which tells us that orbits, the

sizes of orbits are not arbitrary, right, they have to satisfy the
property that they divide the order of the group. So this reduces
the choices and helps us compute the orbits in many situations,
this is the counting formula.
(Refer Slide Time 19:14)

368
So I wanted to do one more application of orbits partition the set
S, right. So let us apply this orbits partition the set S, so assume
now that S is finite. Everything that we have said until this point
the set S need not be finite, because even in the counting
formula the size of S is irrelevant only the size of orbit is
relevant. So in particular I should say that the counting formula
says that if G is a finite group acting on an infinite set S, the
orbits are still finite, so that is a consequence, orbits cannot be
infinite because orbit size must divide the orbit, the group size.
The set S may be infinite but the orbits must be finite.

But now I am going to assume that set is also finite. Then we

have S must be a disjoint union of orbits because orbits are
equivalence classes under the equivalence relation that I defined
at the beginning of this video. So they are of the form Os1 union
Os2 and the symbol for disjoint union is this, denotes. Okay, this
symbol simply means you are taking the union but in addition it
says that the sets involved are disjoint, right.

Because S is a finite set, obviously there are only finitely many

orbits and if you take the distinct ones among them they are all
disjoint and they cover S. So in particular we have the order of
the set S is equal to order of the orbit S1, orbit S2, orbit Sk, this
is another important counting formula, this is also related to the
counting formula. It says that if S is a finite set, then S can be

369
written as sum of orbits.

The cardinality of S, number of elements of S, can be written as

sum of some orbits and because of the counting formula that I
did in the previous slide, each of the numbers here is a divisor of
the cardinality of the group.

So we have a very strong restriction here, so the size of the set S

is the sum of some numbers. But each of those numbers divides
the order of the group. So we are going to exploit in later videos
this is very important observation that we made.

So let me recall, we have made two observations in this video.

One is that if G is a finite group acting on a set S, and we fix an
element small (s) of capital (S), that is very important. You fix,
this is for a fixed element small (s) in capital (S), the stabilizer
order times the orbit order is equal to the, product of the
stabilizer and orbit is equal to, the cardinality of the group itself.
This in particular says, as one consequence, that the number of
elements in an orbit must divide the order of the group.

And another way of using the partition of S into orbits is that if

S is a finite set the number of elements of S is equal to the sum
of number of elements in orbits and each orbit size divides the
orbit of the group.

Let me stop the video here. In subsequent videos we are going to

exploit these counting formulas that we proved in this video,
thank you.

370
371
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 06
Lecture 37 – “Cayley's theorem”
PROF.KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

In this video we are going to prove one of the important

theorems in group theory called Cayley’s theorem. Let me write
it down and I will prove it, it’s an important theorem because it
tells you something about structure of groups and the proof is
fairly easy given that we have understood already some facts
about group actions, ok.

Cayley’s theorem is the following. So let G be a finite group of

order n, that means G has an exactly n elements. Then it says,
then G is isomorphi c to a subgroup of Sn, okay. So it says
simply, this is the theorem: if G is a finite group order n G is
isomorphi c to a subgroup of Sn. Recall that Sn is the symmetric
group on n letters.

So I said this is a structural result about groups because

symmetric groups are very general in this sense, every group
finite group can be put inside in a symmetric group. So if you
study symmetric groups and all their subgroups you would have
studied all groups, that is the point of Cayley’s theorem. Every
finite group is isomorphi c to a subgroup of Sn where n is the

372
order of the group.

So for the proof we are going to consider the action of G on

itself by left multiplication, ok, so when I defined actions, group
actions on a set S, I gave you several examples, two of those the
examples where group acting on itself, so the group is G, the set
is also G in these examples. And there were two important
actions that I highlighted at that point of a group on itself, one is
by left multiplication, and the other is by conjugation ok. So in
this proof we are going to apply the action of G on itself by left
multiplication.

This means if you have G, so here the set is also G and then gs is
simply gs, which is the, gs is simply the product of g and s in G,
right. So this the product inside the group G, so let’s use this.
Now how do I use this? So I am going to fix g in G, first fix
small g in G.

Consider the function G to G, because this function is dependent

on g, let me call this phi sub g, consider the function defined by
I will take small s, phi g of s is gs, so remember G is playing
two avatars here one is as a group the other is as a set. Here I am
thinking of G as a set, I am fixing a group element g, I am
giving you a function on the set to itself, its a function from the
set G to the set G. Because it is determined by small g, I am
going to denote the function by phi subscript g.
So what does it do? It takes an element small s and maps it to gs.
We claim that phi g is a bijection. See certainly phi g is a well-
defined function right, it takes an s and maps it to gs, so there is

373
no problem phi g is a well-defined function.

Now let’s take this function and we want to show it’s a bijection
so in another words we have to check it’s 1-1 and it’s onto. So
why is it 1-1? What is one 1-1? 1-1 means if two things so it’s a
function G to G, if s1 and s2 map to the same thing then s1and
s2 must be equal. So let phi g of s1 equals phi g of s2, suppose
that happens.

This implies, what is phi g s1? that is gs 1, what is phi g of s2?

that is gs 2 gs 1=gs 2. Remember g is the element of capital G s1
is also element of capital G but s1 is being thought of a set
element of the set G so I am using small g to denote the group
element and small s1 and small s2 to denote the element of
capital G when it is being considered as a set. gs 1=gs 2 but this
is a product in the group, that means s1=s2.

The cancellation property in a group, so s1=s2, so phi gs 1=phi

gs 2 means if two things go to the same thing, they are equal to
each other. This shows that phi G is 1-1, right. Two distinct
elements go to the same thing means they are equal.

So now it remains only to show phi g is onto .Why is phi g

onto? What is the meaning of being onto? It means you give me
any element of the target space there exist s something that maps
to it, right. You are given this, there exists such an element. So
now let me give may be given an element s in capital G, so
which element maps to this? So then phi g of g inverse s, let’s
consider g inverse s, right. If s is in G, g inverse s is also in G.

374
So what is phi g of g inverse s? By definition this is g times g
inverse of s, by associability this is g(g inverse)=g(g inverse
s)=(g g inverse)s which is s, right. So if you give me an arbitrary
element of s, g inverse s maps to this under under phi g, so phi g
is onto. So phi g is 1-1 and onto so phi g is a bijection, right.

So this is the claim we made, we claimed that phi g is a bijection

which I proved. G has n elements right, that is the hypothesis,
the hypothesis is that G is a finite group of order n, that means it
has exactly n elements. What is Sn? Sn I said is the symmetric
group on n letters.

So Sn is the group of bijections of this set {1,2,…..,n}, right. But

really the fact that we are calling the elements 1, 2,…, n is not
important, Sn is the group of bijections of a set with n elements,
remember Sn is the group of bijections of an n element set, for
the convenience when working with symmetric groups it is
convenient to call the elements 1, 2, 3, up to n, but it need not
be those, Sn is the group of bijection for any element, n element
set. G is one such, G is an n element set, so Sn can be identified
with bijections of G, because Sn is a group of bijections any set
with n elements and G is a set with an elements, I can naturally
think of Sn as a group of bijections.

So it cannot be identified with bijections of G, I should write can

be identified with the group of bijections ok, so if you wish you
can call the element of G as 1, 2, up to n because it has n

375
elements you fix some order take the first element call it 1
second element is 2 right that’s what we are doing here. Sn can
be thought of bijections of G but phi g is a bijection of G, so
now we have a function, lets call it capital ø from G to Sn, what
does it do? It takes a g and sends it to phi g that I defined earlier,
remember what is phi g for a fixed small g, phi g is the function
from G to G and sends s to gs, which I have checked is a
bijection so phi g we already checked that phi g is a bijection of
G, so phi g is in Sn because Sn is for the purposes of this
theorem we are thinking of Sn as a group of bijections of G
itself, so phi g, being a bijection of G, is an element of Sn. So I
am sending g to phi g. Now ø, capital phi a function from a
group to another group, so now I claim that phi is a group
homomorphism, ok.

So this is, you have to be careful here, so what is the function

phi? It sends an element to another function, the function sends
an element to another function, that’s what capital phi is doing
because Sn is really functions of G to G and a small g under the
function capital phi goes to the function small phi g. So capital
phi is a function that takes an element of the group G and sends
it to a function which is what we are calling by phi g.

So I am now claiming that phi is a group homomorphism, that

is, capital ø(g1g2) =ø(g1) composed with ø(g2). And remember
the composition on Sn when Sn is being thought as bijections of
capital G, the binary operation there is the composition, so you
apply, you take two elements of capital G, g1 and g2, we want to
first multiply them and apply phi then phi of g1 as a function

376
composed with phi of g2 will give you the answer, so this must
be true for all g1, g2 in G ok.
(Refer Slide Time 13:09)

So let’s check this. So now we are saying that two functions are
equal, because capital ø(g1 g2 ) is a function. We are saying
that, ok first let me say, the left hand side LHS and the right
hand side RHS are both functions right, what are they really,
they are functions from G to itself, to G because they are both in
Sn which we are considering as all functions bijective functions
from G to G for the purpose of this theorem. Sn is all bijective
functions from G to G, this we can do right because G has n
elements, so this is valid and now LHS and RHS in this equation
are both functions from G to G.

So in order to be equal, so we are saying that these two functions

are equal, identical as functions. When do we say two functions
are equal? Two functions are equal if they agree on every
element of the domain which this case is G. So let us choose an
arbitrary s in G, again G here is really playing the role of the set,
so I am taking an element small s, let’s take an element s in G,
see what LHS and RHS do to s.

(Refer Slide Time 15:30)

377
Let’s do LHS is ø(g1g2), that’s a function from G to G right,
what does it do to (s)? What does it do to s? ø of (g1g2)
remember is phi (g1g2), because ø as a function from G to Sn
sends a small g to phi g, that’s the function capital ø, so ø(g1g2)
is ø of small phi of (g1g2) and then apply to s, it is this. But
what is this? Small phi was defined way back in the beginning
of the proof.
Small phi of g is (gs) applied to s, so, small (ⱷ) g1, g2 of (s)
g1g2s. This is what LHS does to small (s). Now what does RHS
do? What is RHS?

(Refer Slide Time 16:43)

RHS is (ⱷ) g1 circle (ⱷ) g2, this composition applied to (s),

composition of functions says that this must be (ⱷ) g1 applied to
(ⱷ) g2 applied to (s), that is the definition of composition. But
what is (ⱷ) g2? (ⱷ) g2 is small (ⱷ) g2 (s), right, capital (ⱷ) g2 is
small (ⱷ) sub g2. This is (ⱷ) g2, g1 by the way, here small (ⱷ)
g2 of (s) is by definition of small (ⱷ), is g2 (s). But this is equal
to (ⱷ) g1 of g2 (s), and this is equal to g1g2s. Now are these
equal? Of course they are equal because this is the associativity
in the group, so LHS=RHS as functions. Remember we are
working with an arbitrary(s) so and LHS and RHS agree on that
arbitrary (s), so they are equal as functions.

So this proves that capital (ⱷ) is a group homomorphism, so this

justifies that statement that capital (ⱷ) is a group
homomorphism. Now I want to claim that it is not only a group
homomorphism, it is an injective group homomorphism.

378
(Refer Slide Time 18:27)

So I want to say (ⱷ) of g is 1-1, (ⱷ) is 1-1. Earlier we have

showed that small (ⱷ) is 1-1but remember now we are saying
capital (ⱷ) is 1-1, small (ⱷ) was in fact 1-1 and onto and hence
became an element of symmetric group. Now I am saying
capital (ⱷ) is 1-1, it is generally not onto, it is only 1-1. Why is it
1-1? Because it is a group homomorphism capital (ⱷ) is a group
homomorphism we checked already, it is enough to check what?
A group homomorphism is 1-1 if and only if the kernel of
capital (ⱷ) is the identity element of G, remember a group
homomorphism is 1-1 if and only if its kernel is identity, so this
what I am going to check. So suppose, so clearly (e) belongs to
kernel (ⱷ), remember this requires additional statement, because
this follows immediately from the fact that (ⱷ) is a group
homomorphism.

So it sends identity to identity so, we need to check the opposite

inclusion. So suppose g belongs to kernel (ⱷ), kernel (ⱷ) is a
subgroup of (G), let us take a small (g) in it.

(Refer Slide Time 19:59)

This means (ⱷ) of (g), which is what, small phi of g, remember

capital (ⱷ) is defined to be small (ⱷ) sub (g). So this is, small (g)
is in the kernel means its image is the identity element of S n.
What is the identity element of S n? Sn is the group of bijections

379
of G, it is identity element means it is identity function. So, (ⱷ) g
is the identity function from G to G. Identity element of the
symmetric group is the identity function. So (ⱷ) g is the identity
function. Hence, phi g of (s) =s for all s in S, this is the
conclusion of g being in the kernel of capital (ⱷ).
(Refer Slide Time 21:16)

But what is phi g of s? ⱷ g of s is by definition (gs). So gs=s for

all s for all s in S. In fact, this is a very strong condition, gs=s
remember implies (gs) s inverse=s s inverse implies g=e, this is
the cancelation property, right. We can cancel s, that is the
point.

So, g=e, isn’t that what we wanted? We said that kernel is

identity only, we started with an arbitrary g in kernel (ⱷ) and we
concluded that it is equal to e. So, capital (ⱷ) is1-1.

(Refer Slide time 22:10)

Now lets us take stock of where we are so, now situation is,
capital (ⱷ) from G to Sn is a 1-1 group homomorphism. It is a 1-
1 group homomorphism. Now, we recall a point that I made
after proving the first isomorphism theorem, if you have a 1-1
group homomorphism or injective group homomorphism, G can
be identified with a subgroup of Sn.

380
So by the first isomorphism theorem, remember first
isomorphism says that a group homomorphism if you have G
mod the kernel is isomorphic to the image. So G is, in this case
the kernel is trivial right because it is injective, to the image. By
the first isomorphism theorem G is isomorphic to the image.
(Refer Slide Time23:13)

But note that the image of (ⱷ), in general the image of a group
homomorphism is always a subgroup of the target group. In this
case it is a Sn so image of (ⱷ) is a subgroup of Sn and G is
isomorphic to it. So, this completes the proof of Cayley’s
theorem.

Remember what was Cayley’s theorem? So that was the

beginning of this video. Cayley’s theorem said that if you have a
finite group of order n, then it isomorphic to a subgroup of S n ,
which is exactly what we have just proved.

G is isomorphic to the image of (ⱷ) and image of (ⱷ) is a

subgroup of Sn , so G is isomorphic to a subgroup of Sn so, this
is Cayley’s theorem and this is a structural statement, not always
very useful and the reason that it’s not always very useful is
because Sn is a very complicated group, okay.

So, one thing that we have noticed in that video when we talked
about Sn is the size of Sn keeps increasing by larger and larger

381
numbers because S3 has order six, S4 has order 24, S5 has order
120, S6 has order 720 and so on. So S n is a very complicated
group, so knowing that every group is a subgroup of S n is not in
practice very useful, however it is a good structural statement.

It says that every group can be thought of as a subgroup of S n.

So I am going to remark here something that we have proved
earlier.

(Refer Slide Time 25:15)

Re call that a cyclic group is isomorphic to a quotient of the

form. So compare this statement, cyclic group is isomorphic to
quotient of the form. So, here also we are relating an arbitrary
cyclic group to a well-known group like Z. So how do we show
this?

We take the map from Z to G sending 1 to the generator of G.

Because G is a cyclic group this is onto and kernel is always of
the form NZ for some N, for some N, I should write here. So, by
the first isomorphism theorem Z mod NZ is isomorphic to G. So
this is also a structural theorem, we are relating an arbitrary
cyclic group to something well known.

Cayley’s theorem does the same. It relates in fact an arbitrary

group any finite group not just a cyclic group, any finite group is
actually can be put inside Sn, think of it like this, it can be, it is
isomorphic to subgroup of Sn, so that means you can put it
inside Sn.

382
But the reason why this is very useful? That cyclic groups are of
the form Z mod NZ is that Z is very simple group to
understand, so Z mod NZ is a very simple group and to say that
every cyclic group and to say that every cyclic group is of that
form is very nice.

Whereas Sn is a very complicated group, so knowing that every

group is a group of Sn is not in general that useful, okay.
Nevertheless this is an important theorem in group theory
because it says that every group sits inside Sn. So in theory you
know that subgroups of Sn exhaust all finite groups, that is what
it means. If you list all the subgroups of the symmetric groups
for all N, then you have listed all groups.

And that is a strong theoretical statement, and it is sometimes

useful to, if you work with an arbitrary group G, it sometimes, to
give it a concrete shape, and you can do that by putting inside
Sn, and we have done a lot of analysis of Sn, it has a cycle
decomposition, it has every element can written as a product of
disjoint cycles, every element is the a product of transpositions,
there is even and odd permutations.

So now knowing that every group is isomorphic to a subgroup

of Sn, you can really think of elements of G as cycles, because G
is sitting inside Sn and as far as group-theoretic properties are
concerned, they are preserved under the isomorphism, so G can

383
be replaced by whichever group it is isomorphic to which is a
subgroup of Sn and work with whatever we know about Sn. So in
some cases it is a good result, because it helps us give a concrete
shape to an abstract group. And that is why Cayley’s theorem is
an important theorem in group theory, so let me stop the video
here in the next video we will continue our study of group
actions, thank you.

384
 Introduction to Abstract
Group Theory
Module 07
Lecture 38-“Problems 7”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHAMATICAL INSTITUTE

In this video let us do some problems on group actions and

stabilizers and orbits.
(Refer Slide Time 00:22)

So I want to do some problems in this video. Let us start with

the following. So let us start with some calculation of stabilizers
and orbits, okay. Let us consider the action of, let us take G,
consider the action of a group G on itself by left multiplication.
We know from viewing the action as an equivalence relation, we
know that the group G, the set S when a group G acts on a set S,
S is partitioned into orbits, S is a disjoint union of the orbits.

Find the orbit decomposition of G, and similarly find the

stabilizer of an element s in G. This is the problem, because G is
partitioned as a disjoint union of orbits, so I want you to find all
orbits of G and see what are the orbits whose disjoint union is G.
Similarly given an elements small s find it stabilizer. So recall
that here the action is given by left multiplication g, s maps to
gs, so this the action, lets compute the order.
So let us first take compute the orbit of an element, in order to

385
find the orbit decomposition, let us start with finding the orbit of
a single element. So let us take s in S, what is the orbit of s? So
orbit of s is gs as, by definition it is the set of the elements gs, as
g varies in G.

Now this is something that I have implicitly done several times,

but the point is this is equal to G, so of course the orbit is
contained in G always, G being thought of the set here, but in
fact orbit is equal to G. Because if t is in G, then g lets say s t
inverse, sorry if t is in G, we want to find a group element such
that g times s is equal to that, so what should that be? So how do
we check this? We want gs is to be t, if t is in G, I want to show
that t is in Os, but OSs consists of gs, so gs must be equal to t is
for some g, so I want to solve for g, so g would be t(s inverse).

So we take t(s inverse) times s is t. So that means t is in Os, so

orbit is G and this is remember for all s in G. So what is the orbit
decomposition? So there is exactly one orbit for this action, in
other words all orbits are identical to G, so this such an action is
called “transitive action”, so I will define this may be formally
again in some other video.
(Refer Slide Time 4:56)

But this is an example of transitive action, transitive means only

one orbit, so this is the orbit decomposition, so we have done the
first part of the problem, write the orbit decomposition so G is
equal to Os, that is orbit decomposition, for any s. It is in general
a disjoint union of orbits but there is only one orbit here.

386
Now on the other hand, let us take an element s in S in this case
G, find the stabilizer. That was also the problem right, what is
Gs? This is all elements of group now that fix s, all elements of
group that fix s, what are these? If gs is equals s, a quick
calculation tells you, in this also we have done earlier, so the
stabilizer of any element is just {e}, okay.

And this illustrates the counting formula that we have done, so

think for a moment that G is finite, remember that for any action
we have, recall the counting formula .
(Refer Slide Time 6:54)

Counting formula says that order of the group is equal to

number of elements of the orbit times the number of elements of
the stabilizer, for a fixed elements, here again S is an arbitrary
set and G acts on it. So remember the dynamics here, this left
hand side is fixed, if for some element there is a large orbit, the
stabilizer must be small, if the orbit is small the stabilizer must
be big, because the product of these two is order of G. So large
orbit, loosely speaking, this implies small stabilizer. Small orbit
implies large stabilizer correct, small orbit means Os is small
number of elements, so Gs must make up for it because the
product is order of G, so very loosely speaking this what you
this is a heuristic that you should remember.

In this case for the group action given by left multiplication,

orbit is very big. As big as it can be, so stabilizer is very small,
as small it can be, one element all elements, so that is what

387
happens in this problem.

Let us do problem 2, do the same for the action of G on itself by

conjugation when G is abelian. So let us assume in this problem
that G is abelian, let us find out orbit decomposition and its
stabilizers for this action. What are these? So now remember the
action is given by gsg inverse, that is the action, conjugation
action.
(Refer Slide Time 9:23)

So what is the orbit of an element? So orbit of an element, lets

us fix s in S or s in G, what is the orbit? This is gsg inverse as g
is inside G, but I am assuming G is abelian so that means gsg
inverse is just s, so no matter what small g, this is just s. And
what is the stabilizer? So here the reverse happens, as opposed
to the pervious example problem. Here orbit is very small, orbit
is as small it can be, it only consists of that element.

On the other hand, what is the stabilizer? This is g in G such that

gsg inverse is s, things that fix s under the action gsg inverse is
the action of small g and small s, that must be equal to s. But
this is g in G such that s equals s, but there is no condition, in
other words for all g in G this condition holds, so this is G.

So here orbits are small, stabilizer is big, stabilizer is as big as it

can be, it is all of G. So orbit decomposition would be G would
be union of singleton elements, earlier there is only one orbit in
the previous example, so there is just one orbit, here there are as
many orbits as there are elements of the group, so this the orbit

388
decomposition.
(Refer Slide Time 11:13)

Let us do problem 3. Show that so now again consider the action

of G on itself by conjugation. So I am not going to assume now
that G is abelian, G is an arbitrary group. Show that, two things,
so let s be an element of G, show that stabilizer of s is the
centralizer of s. I will recall in this problem what is the
centralizer, so stabilizer of s is the centralizer of s and show that
s belongs to the centre of G, I will also recall for you what is the
centre of a group, if and only if stabilizer is all of G. So this the
problem. So let us solve this, so I am asking you to show that the
centralizer of s.
(Refer Slide Time 12:22)

So I am going to recall, let G be an aribtrary group, and let us

take g in G, centralizer of g, which I have denoted by C of g I
think, is all elements which commute with g, so a in G such that
ag equals ga. Remember this is same as saying that a and g such
that aga inverse is g. What is the centre of G, of the group G
which I denoted by ZG, this is elements of the group that
commute with everything else. This centre will always be
contained in the centralizer of any element. So this is all
elements such that aga inverse is a for all a in G, for all g in G,
right.

So this is the centralizer and the center that we have defined

389
earlier. Now for the action of conjugation, action of G on itself
by conjugation show that the stabilizer is the centralizer, now
what is the stabilizer? This is almost clear from the definition
now, right, because Gs is by definition g in G such that gs g
inverse is s, the stabilizer of s. But this is of course the
centralizer as I have written here, if you just stare at the two
things, it is all elements such that when you take the conjugate
of g or this particular element, you get that element back, so I
am of course confusing with the letters here, here am using a
and g, here am using s and g but hopefully it’s clear that
stabilizer of an element s, is all g such that conjugate of s by g is
equal to s, which is the centralizer of s. So this the first part of
the problem.
(Refer Slide Time 14:57)

Now suppose, the second part of the problem is, an element is in

the center if and only its stabilizer is G. Suppose s is in the
center, if s is in the center, this means sx, rather I should write sg
= gs for all g in G, this is the meaning of being in the center
which is same as gsg inverse is s for all, these are all if and only
if, right, they go back and forth. If s is in the center this
happens, if this happens then s is in the center. This is clearly
equivalent to this. But this means g belongs to stabilizer of s for
all g in G, right. If gsg inverse is equal to s, that means the g is
in the stabilizer so if this gsg inverse is s for all g then g is in the
stabilizer for all g. Similarly if g is in the stabilizer, gsg inverse
is s. But if g is in stabilizer for all g, that is equal to this, so s is

390
in the stabilizer if and only if gs =g, this the second part of the
problem, that also I have shown okay.
(Refer Slide Time: 16:38)

So now let us look at a different action, so a different example.

So let’s look at problem 4. Here I am going to consider the
action the group of invertible n by n matrices on the set of
column vectors, here the action is given by A.V is AV, where
AV is matrix multiplication okay. So here again find the orbit
decomposition of R^N, and two find the stabilizer of the
elementary vector E1, so that is (1, 0, ..,0), so two problems
okay.
(Refer Slide Time: 17:51)

Let us solve this find the orbit decomposition, so I have alluded

to this when I did this example and did orbits, so let me quickly
do this. First of all, if you take the 0 vector, the orbit of the 0
vector is just the 0 vector right, because any matrix times 0
vector is, this is a 0 vector, so this an orbit by itself. So this is
one orbit. On the other hand, there is only one more orbit,
because given any nonzero vectors V1 and V2, there exists an
invertible matrix A such that AV1 is V2.

So this something that I said is a fact, that I don’t want to prove

this now. But you can check easily at least when N =2, so check
this, I will leave this as an exercise for you for N = 2. It is very
easy to check this because it is all 2 by 2 matrices and vectors of

391
length 2, you can easily cook up a matrix which achieves this,
given V1 and V2.

That means all nonzero vectors form an orbit, right. Is that

clear? Because I am saying that any two vectors given any two
nonzero vectors there is a matrix which sends one to the other,
in other words any two nonzero vectors are equivalent.

So all nonzero vectors together form an orbit. So what is the

orbit decomposition of RN? You have the 0 vector as one orbit,
so there are, and all nonzero vectors as one orbit. So there are
exactly two orbits for this action.

Note that all counting formula and so on don’t make any sense
here because the group is infinite and the set is infinite, so we
cannot get anything from counting formula. However we can
say that the set is a disjoint union of orbits, that always holds
irrespective of the finiteness of the group or the set. And in this
case the two orbits are 0 vector by itself and nonzero vectors by
themselves. So this is the orbit decomposition modulo this fact
which I will leave for you the verify, in general also it is not
difficult and N =2 is especially easy.

(Refer Slide Time: 20:46)

Now let us do the second part. What is the stabilizer of E1, so
this is G(E1) is my notation for this. So these are all matrices in
GL_N such that A E1 is E1. Let us work it out a little bit.

Let us write a matrix like this, it is an N by N matrix, so first

entries, row are A11 and A12, A22 A2N finally the last row is

392
AN1 and AN2, ANN. And if you multiply this with E this is A,
E is (1, 0, 0,…,0). What do you get?

If you are familiar with matrix multiplication, this is simply A11

times 1 + A12 times 0 and rest do not contribute so that is just
A11. The second row times the vector here will give you A21
upto AN1, so this is the first row, first column rather, okay.
This is a property of matrix multiplication, when you multiply
the matrix with the E1 elementary vector you get the first
column. But what is the, if A is in the stabilizer then AE1 is E1.
That means the first column of the matrix A must be (1, 0,
0,…,0), E1 itself. And there is no other condition, right. So
stabilizer of E1 or G(E1) is matrices such that the first column of
A is E1, okay.

More clearly, it would be all matrices of this form, the first

column will be A and the rest is anything, this part is arbitrary.
Of course, it must be something in GLN(R), right. So stabilizer
of E1 is, because remember there is no condition on A12, A1N,
A22, A2N, AN2, ANN, they do not contribute to the product at
all, so they can be anything.

Only the condition is first column must be nonzero, must be E1,

this is arbitrary, star means it can be anything, but in a way that
this whole matrix forms an invertible matrix. So this is the
stabilizer of E1. So this is the problem, right: we have shown
that orbit decomposition of this is the 0 vector union nonzero
vectors, and we have computed the stabilizer of E1.

393
 Introduction to Abstract
Group Theory
Module 07
Lecture 39 – “Problems 8 and Class equation”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

(Refer Slide Time: 00:16)

Let us do a related problem. I will write it as a separate problem,

and problem number would be 5. Problem 5 is, for the same G
as GLn(R) and S as Rn as in problem 4, okay, show that if A
satisfies, if A is a, matrix invertible N by N matrix such that A
dot V is V for all V in Rn, then A is the identity n by n matrix.
Why is this? Okay, so any matrix which fixes every vector, so in
other words, it is in the stabilizer of all vectors, all elements of
the set S, then it must be the identity element.
Why is this? So we already saw that Ae1 equals to e1 implies,
the first column, right, this we have seen already in the previous
problem. And Ae1 is equal to e1 because AV is equal to V for
all V. So now Ae2 is also equal to e2, right. e2 is the second
elementary vector. So 0, second entry is 1 then followed by 0s.
This implies so A is already we know this but second column
can be anything at this point.

So this is a(12), a22, a 2n, a1n, a2n, ann, multiplied by 0,

1,0,…,0 must be 0, 1, 0,…, 0. And exactly the same calculation
as before shows that, this implies, what is this product? This
picks out the second column of the matrix. Because when you

394
multiply it out, the second column is picked out ,this is
0,1,0,0,….,0. That means if Ae2 is e2 the second column of A is
e2. Right, we already know the first column of Ae is e1. Now
we know Aei is equal to ei, so the ith column of A is ei. So, now
this is true for all numbers from 1 to n.

So the first column of is A is e1. Second column is e2, third

column e3, fourth column is e4, nth column is en, but this means
A is identity. Right, so A fixes, in fact the problem asked you to
show that if AV is equal to V for all v in R^n, then A is identity
matrix. It suffices to show, actually assume that if Ae1 is e1,
Ae2 is e2, …, Aen is en.
(Refer Slide Time: 04:22)

But of course if you think about it a little bit, if Aei is equal to ei

for all i, AV will be equal to V for all V. This is because the
vectors e1, e2, …, en span the vector space R^n, if you know
this, but it is not important for this problem.
(Refer Slide Time: 04:49)

So this action has the property that if a vector, if a group element

is in the stabilizer in the all vectors, then it is the identity
element. So this action is an example of “faithful action” okay.
So I will define this now which I have not defined earlier. So
definition: Let a group G act on a set S. We say that the action of
G on S is faithful, we also say sometimes, or
(Refer Slide Time: 05:35)

395
“G acts faithfully on S”, so action is G on S is faithful or G acts
faithfully on S, if g is an element such that gs is equal to s in S
this implies that g is the identity. Right, so if g is an element of
the group that fixes all elements in the set then g is the identity
element okay. So this is an important class of actions, so G acts
faithfully on S. So other examples include, action of G on itself
by left multiplication is faithful.

And this is exactly what we used in the proof of Cayley’s

theorem. On the other hand, action of G on itself by conjugation
is not faithful in general. So for example, if G is abelian, then
orbit of Gs is G, that is, the stabilizer is G, for all s, as we saw
earlier. The orbit is G right, this is one of the problems we did
earlier. So certainly, the intersection of the stabilizers is all of G.

On the other hand, this is a problem that I will eave for you as an
exercise, I won’t do. But it is an immediate consequence of the
problem on the relevant problem here. So problem 6.
(Refer slide Time: 08:06)

Show that the action of G on itself by conjugation is faithful if

and only if the center is trivial. Okay, the center I have defined
also in the previous problems. See if G is abelian it is very far
from being faithful. It is in fact the diametrically opposite to
being faithful because faithful means intersection of orbits is
trivial, sorry intersection of stabilizers, I keep confusing between

396
orbits and stabilizers, but faithful means intersection of
stabilizers is trivial. But if it is abelian all stabilizers are the
group G, so certainly it is not faithful.

So faithful and abelian are at the opposite ends of the spectrum

as for as conjugation action of G on itself is concerned. So
faithful and opposite spectrum of abelian is the center being
trivial. Center being the whole thing is abelian, center being
trivial is opposite thing, so if the center is trivial, the action of G
on itself by conjugation is faithful. Let us finally do one more
example. This is computational, to compute equivalence classes
and orbits.
So consider, let G be S3. Compute the orbits for the action of S3
on itself by conjugation. Remember action of S3 on itself by left
multiplication, orbits are only one. There is only one orbit. So
that is not interesting, and stabilizers are all trivial okay, so
operation, or action of S3 on itself by conjugation is more
important, more interesting.
(Refer Slide Time: 10:28)

So let me use this representation of S3, cycle notation, so (12),

(13), (23), (123), (132), okay. Let us use the, so a quick piece of
definition. Orbits for the action of G on itself by conjugation are
called “conjugacy classes”.

Okay, so this is an important notion, so this is exactly as orbits,

it is just a new word. Conjugacy classes is the word in the
specific example of the action of a group on itself by

397
conjugation okay. So this is useful to, because the action of a
group on itself by conjugation is an extremely important
example of group action.

And this is very important in proving various important

theorems, so the orbits are given a special name in this case.
They are called conjugacy classes.

So the problem 7 is simply asking you to find conjugacy classes

of S3. That is just shortened version of saying find the orbits for
the action of S3 on itself by conjugation. In other words, find the
conjugacy classes of S3.
(Refer Slide Time: 12:04)
Okay, so what is the orbit of e? It is simply, remember orbit of e
for the conjugation action is just e, because we have to do geg
inverse, as g is in G. But this is just e, right. The action is by
conjugation, so you vary small g and take geg inverse, this is
okay.

Now find the orbit of let say (12). Wo we want to do g (12) g

inverse as g varies in S3. Okay, so this is a quick calculation, so
let us do one by one.

So e(12)e = (12), that is always there. So (12) is always in the

orbit. What about (12) (12)(12) inverse, what is inverse of (12),
so this is (12)(12)inverse again(12). This is e, that (12),(12) is
identity, so that is not surprising. So the new thing will be if you
take (13). So do (13),(12)(13) inverse, so what is (13) inverse it
is again (13), so I have to vary g and compute g(12)g inverses
and what is (13) (12) (13). If you do the calculation quickly (13),

398
1 goes to 3, 3 goes to 1 so 1 goes to 1, 2 goes to 1 and 1 goes to
3, and 3 goes to 1, 1 goes to 2, so 3 goes to 2.

Similarly (23) (12) (23), because inverse of (23) is again (23)

will be I will let you calculate this quickly, it is (13). Because 1
goes to 2 and 2 goes to 3, 3 goes to 2, 2 goes to 1 so 3 goes to 1,
2 goes to 3, 3 goes to 2, okay. So now let us do (123)(12),
inverse of (123) is (132), if you think about it for a minute. So
this would be 1 goes to 3, 3 goes to 1, so that is 1; 2 goes to 1, 1
goes to 2, 2 goes to 3; 3 goes to 2, 2 goes to 1, 1 goes 2; so it is
(23). And finally (132)(12) and (132) inverse is (123), and this
will be 1 goes to 2, 2 goes to 1, 1 goes to 3; 3 goes to 1, 1 goes
to 2, 2 goes to 1; 2 goes to 3, 3 goes to 2; so this is (12). So what
is the conjugacy class of (12) or the orbit of (12)? This is equal
to, (12) certainly will be there, (23) is there, and (13) is there.
So, in fact, we get all 2-cycles or all transpositions, right this is.
(Refer Slide Time: 15:16)

So now recall the splitting of S as a union of disjoint orbits, so

S3 remember must be a disjoint union of conjugacy classes,
disjoint union of orbits, in this example we are calling them
conjugacy classes, and so far we have found two of them. So, so
far we have found e that is one conjugacy class and all two
cycles form another conjugacy class, so this is one conjugacy
class, this is second conjugacy class so, so far we have managed
to account for four elements of the symmetric group on three
letters.
The remaining two elements must be also part of some
conjugacy classes. So find the orbit of (123), Let’s say. So in

399
order to find the orbit of (123), let’s find the conjugate of (123)
by (12), (12) inverse as before is (12). So (12)(123)(12), let’s
compute this. 1 goes to 2, 2 goes to 3, so 1 goes to 3; 3 goes to 1,
1 goes to 2 so (132); 2 goes to 1,1 goes to 2, 2 goes to 1 so (13).
So this means (123) is related to (132) and now you are able to
conclude that so we can conclude that (123) and (132) is a
conjugacy class, right, because so remember we have to find
one more. That this must be the conjugacy class, why is that?
These are already part of the conjugacy class, (123) and (132)
are related.
So they are part of a conjugacy class, but can there be anything
else in this conjugacy class? No, because S3 only has six
elements and the remaining four elements are already accounted
for, if any of them are in this conjugacy class this conjugacy
class will have a nonempty intersection with them but conjugacy
classes need to be disjoint so this is the conjugacy class and you
can also check direct by calculation that all other conjugates will
be equal to one of these but we do not need to do that.
(Refer Slide Time: 18:01)

So the conjugacy class decomposition of S3 would be e disjoint

union with (12), (23) and (13) disjoint union with (123),(132).
Now let’s observe one thing. When I talked about counting
formulas I noted that the cardinality of the set, of the group or of
the set, which in this case is again the group, is the sum of
cardinalities of orbits, right. And I also noted at that time that
cardinalities of orbits divides the order of the group, so in this
cases what is that formula this is called the “class equation”
okay.

400
So in the case of a group, a finite group, I will define this more
formally later when I come to this later in a video. In the case of
finite group acting on itself by conjugation this statement that
cardinality of S or the cardinality of the group which is also the
set in this same example, is equal to order of one element orbit
of it is equal to sum of sizes of orbits, is called the “class
equation” of finite and group G acting on itself by conjugation is
called the class equation.

This is a very, very important piece of data attached to a group

okay, as I said the action of G on itself by conjugation is a very
very important group action and the usual things that we have in
all group actions, are given special names in this case because of
the importance of this example. So, orbits are called conjugacy
classes and this formula here is called the class equation.

So, the class equation of S3 is, so you have 6 is the cardinality

order of S3, one orbit is just containing the identity, that is 1. The
other orbit contains all 3 cycles. There are three of them and the
final orbit contains two 2-cycles. So, this is one element, this is
3 elements; 1 element, 3 elements, and 2 elements.

So, the class equation is 6 equals 1 plus 3 plus 2 (6=1+3+2) and

remember another point that we made earlier checks out. Each
number that appears in the class equation must divide the order
of the group which is 6 in this case. 1 divides 6, 3 divides 6, 2
divides 6. Because these are orbits and the counting formula
says that orbit size divides the orbit of the group. Now using this

401
you can also compute the stabilizers of these elements, because
remember
(Refer Slide Time: 21:25)

orbit of (12) is (12), (23) and (13). So, remember the stabilizer
of (12) or size of that and orbit of (12) size of that must equal 6,
this is the order of S3. And this is 3. So, stabilizer of, must be 2.
What is it? Let us compute this. Stabilizer of this would be,
remember there are some obvious elements in the stabilizer.
G(12) so, I am doing more than what is in the problem. So, really
we have finished the problem.

Problem was compute the conjugacy classes of S3 which I have

done. But let us go and do, exploit whatever we have found, in
more detail. The stabilizer of (12) is elements of S3 such that g
(12) g inverse is (12). Right that means this is all group elements
that commute with 12. So, definitely e belongs to G (12) and (12)
belongs to G (12). Right because e certainly commutes with (12)
and (12) commutes with (12)
(Refer Slide Time: 23:14)

But order of G (12) the stabilizer is 2 and it already contains two

elements, namely e and (12). So, the stabilizer of (12) must be e
and 12. The counting formula tells us that we do not need to
now check for other elements because the stabilizer must

402
contain exactly 2 elements and it contains these two elements, so
no need to check make further checkings. This is the stabilizer
of (12).

Similarly, what is the stabilizer of (13)? It also must be of size 2

and contain e and (13). Similarly, stabilizer of (23) must contain
identity and (23) and must be of size 2. So, this is the stabilizer.
Finally, what is this stabilizer of (123)? The size of the stabilizer
must be six by two because the orbit of (123) is 3 right? Is 2,
orbit of (123) that we calculated has two elements (123) and
(132). So, the stabilizer must have three elements. Stabilizer
must have 3 elements and
(Refer Slide Time: 24:52)

certainly e and (123) are contained in the stabilizer and in fact

since stabilizer is a group, is a subgroup of S 3 and (123) is in it,
its inverse is also there. So, inverse of that is (132), (132) is in
the stabilizer. So G (123) contains e, contains (123) and contains
(132). But it is group of size 3 so, that must be it. This is of
course also equal to the stabilizer of (132). Okay, so this
extended problem here hopefully gave you a clear idea of the
power of the counting formula and the class equation that we
have computed, we have proved in the previous video.
So, again just to recap, the action of a group on itself by
conjugation is a very important example of a group action and
we have worked out in this problem fully all there is to work out
in the case of the symmetry group on 3 letters.

We first computed the conjugacy classes by looking at orbits of

403
elements, there happened to be three orbits. Orbit of e which is
one element, orbit of (12) is 3 2-cycles, orbit of (123) is two 3-
cycles.
So, the class equation which is simply the expression of the
order of the set as the sum of sizes of orbits. In this, it is 6 equals
1 plus 3 plus 2 (6=1+3+2). They are three orbits of sizes 1, 3 and
2, so they add up to 6 and we noted that 1 divides 6, 3 divides 6,
2 divides 6, as we would expect because orbit size divides the
orbit of the order of the group.
And then using the counting formula we guessed or rather we
computed the stabilizers by first noting that stabilizers must
have either orbit or size 2 or 3 depending on what element we
are taking and by picking obvious elements in this stabilizer, we
are able to compute the stabilizer of (12), (13), (23) and also of
the 3-cycle (123). Stabilizer of (123) happens to be the
stabilizer of (132).

Okay, so hopefully all these problems in this video give you

clarity on group actions and the remaining content of the course
would be using the class equation and conjugacy classes to
prove some important results about finite groups called Sylow
theorems which we will do in the next video. Thank you.

404
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 08
Lecture 40- “Group action on subset”
Prof. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

Okay, in this last part of the course we are going to study the
Sylow theorems,
(Refer Slide Time: 00:21)

which is the last topic. Sylow theorems are very important part
of finite group theory, they are standard theorems and they are
very useful in understanding finite groups. Before we state, there
are three Sylow theorems, I will state them and prove them in
next few videos. Let me first recall what we have learned in the
previous videos.

So our setup was a following. So G is a finite group, so we are

going to recall first. So remember our setup is G is a finite
group, S is a set okay, so most of the time we will deal with
finite sets and the assumption is G acts on S, right. So there is a
map from G cross S to S, you have a an element g and an
element s, we map it to what we call g dot s. So sometimes we
just write it as gs. And it has usual properties 1 dot s is s, and
(g1g2)s is g1(g2s).

405
Okay, so these are the properties of group action. So now and
the most important thing that we have learned,
(Refer Slide Time: 01:56)
in the video when we talked about group actions was the
counting formula. Remember, so let say small s is an element of
capital S. So S is the set on which G acts, so then the stabilizer
of s, we have defined stabilizer of s denoted by Gs or sometime I
will denote it by stab(s), and that is all group elements which fix
s or stabilize s, remember that this is a subgroup, subgroup of,
this is a subgroup of G. So this is a subgroup of G and the other
important set that we have attached to small s is orbit of s which
are denoted by Os, it is a subset of s given by gs, g belonging to
G, right. So this is the orbit, this is a subset of S.

These are the two things that we attached to a particular element,

so these are attached to a small s, and the counting formula says
the order of G is equal to order of the stabilizer times the order
of the orbit, meaning number of elements. So you take the order
of the subgroup Gs and multiply with the order of the orbit or
the number of elements in orbit you get the order of the group.
And remember how do we do this, we observed that this
follows, remember this follows because, this follows from the
bijection between, this follows bijection between G mod Gs and
Os.
So we take a coset and map it to Gs. So this was the proof of the
counting formula. And another important property of group
actions is,
(Refer Slide Time: 04:27)

406
equivalence classes, so the action of G on S partitions S into
disjoint orbits, right. So if S is a finite set, then we can write the
order of S or the size of S or the number of element of S as the
sum of sizes of disjoint orbits so let us call this o1, o2, …, ok,
where o1,..,ok are distinct orbits. Remember that orbits, two
different orbits are always disjoint, so the two orbits are distinct,
so they are either disjoint or same.

So you can always use this, these are the two important
equations that we have attached to a group action. We have the
counting formula which says something about the order of the
group and then we have something about the order of set, and
the most important actions for us in the context of the Sylow
theorems.
(Refer Slide Time: 05:54)

So we are interested in the action of a group G on itself.

Remember that the two equations that I recall for you are
general statements, whenever a finite group act on a finite set S
they hold. But we want to focus in these videos on the group G
acting on itself. So in other words the set is also S, the set is also
G and we are interested in the two important actions are the
following.

So one, G acts on G by left multiplication, remember this means

the action is given by g dot let me use the letter s, so g and s are
of course in G but I want to distinguish between the group
element and element when thought of as a set. So this is just

407
action, so multiplication on the left. The other important,
(Refer Slide Time: 07:16)

action on G on itself is by conjugation. So here g acts on s by

conjugation so gs is defined to be gs, g-1. We checked in our
earlier video that these are the actions of G on itself. So this
gives the important class equation, this was the other important
thing that we have learned about group actions. The action of G
on itself by conjugation gives us the class equation. What is the
class equation?

So we can write the order of G as order of conjugacy classes. So

c1 through ck are conjugacy classes, distinct conjugacy classes
of G. Recall that conjugacy class remember what is conjugacy
class? Conjugacy class is simply orbit its an orbit for G acting
on itself by conjugation. Okay, so I won’t go into details
because we have discussed this, except to say that,
(Refer Slide Time: 09:09)

there is a class equation for every finite group, we can compute

in principle class equation of any finite group. I will recall for
you, for S3 order is 6 and there are three conjugacy classes of
orders 1, 2, and 3, so this is the class equation of S3. Okay, so
now this is all for recalling.

Let me now go to the new stuff. So I want to in this video setup

the Sylow theorems, first Sylow theorem and prove it.

408
So for that in these three theorems that we will call Sylow
theorems, we are going to repeatedly and alternately use G
acting on itself by conjugation sometimes, by left multiplication
sometimes.
(Refer Slide Time: 09:59)

So suppose, so now suppose, G acts on itself, consider the action

of G on itself by left multiplication, so one of the first of the two
operations I mentioned of a group on itself. So this is the action
of G left, on itself by left multiplication.

So now I want to define we want to define an action of G on

subsets of S. Okay, so we know that G acts on itself by left
multiplication, using that I want to define an action of G on
subsets of S so S is actually G on subsets of G. So suppose let A
be a subset, so I want to stress here, it is only a subset in general,
not necessarily a subgroup. So it is just a subset. It is some
collection of elements of G. And let small g be an element of the
group.

Define g dot A, the definition of g dot A is very natural, so it is

ga as a various in small a varies in capital A. So I will leave it as
an exercise for you,
(Refer Slide Time: 11:33)

and this is very easy exercise, so if you recall or define the

power set of G. So this is P of G is the set of all subsets of G, so

409
the above action, so what I defined earlier defines an action of G
on the power set. Right, so I define g of A to be ga, g times
small a, as a varies in capital A. So now you can define this for
any subset so I claim that it is an action of G on this new set
namely the power set. What are the conditions to check?

Identity times any subset itself, that is clear, because if g is

identity, g times a is just a and similarly the associativity
property holds, which is trivial to check. So we have in action of
the group on the power set, so now let us fix a subset of G.

What is a stabilizer of A? Let us explore this, what is the

stabilizer of A? Okay, so remember stabilizer is, so I am going
to use this notation, stabilizer of A is all group elements such
that gA is A that’s all.

So stabilizer is this, so I want to stress here that gA equal to A,

means, what does it mean? So the set gA as a is in A, which is
exactly gA is A. So in other words, if a belongs to a then ga also
belong to A. This is the meaning of small g stabilizing capital A,
so it must have the property that if small a belongs to capital A,
then small ga belongs to capital A. Then we say that it stabilizes
capital A, in particular it is not necessary that ga is equal to a,
right that is not necessary, ga is equal to a, need not happen.

We do not need ga to be a, so it is not necessary. We need that

ga is again in A, so certainly g times capital A equal to capital A
doesn’t mean g times small a is equal to small a.

410
So with that, now suppose that, so now let’s go back to the
general situation, G is a group acting on itself by left
multiplication, okay, so now we are in the general situation that
G is a group acting on itself by left multiplication and I have
extended that action to subsets of G. So now let’s look at
following situation.
(Refer Slide Time: 15:52)

Suppose that you have again, G acts on itself by left

multiplication and A is a subset of G, we have extended the
action of G to subsets as above. Okay, now I am going to
consider, let the stabilizer be, let H be the stabilizer of A, so
remember this means it is all elements which fix the set A, not
an element of A, they do not fix an element of A. They are
allowed to interchange elements of A, you take an element of A
small g maps it to some other element of capital A.

Okay, now I am going to prove that, so this, I am going to prove

a small lemma here which I will use later. So the lemma is the
order of H divides the order of A. So whenever I use this vertical
bars I always mean order of that set, meaning the number of
elements of that set. Okay, so and also I must tell you again that
in the all the videos from now on G is a finite group. So we
assumed that G is a finite group, okay so what is the proof of
this?

So I want to show that order of H divides order of A, and the

411
reason is remember H is the stabilizer of A, so let small a be in
capital A. Then since H stabilizers A, ha is in A for every h in H.
So the set in the other words, the set ha as h varies in capital H is
contained in capital A,right. So the, I want to call like this, I
want to write like this, the H-orbit of a is completely inside A, is
that clear? The H-orbit of small a is completely inside A. What
is H-orbit?
H-orbit is simply what I wrote earlier. So ha, small ha, as small
h in H. H-orbit is simply the orbit of A as you take elements of
capital H, I am calling H-orbit as opposed to orbit, because I am,
now I have a different group right, earlier I am considering G-
action now I am considering H-action, so orbits are dependent
on the group which is acting.
So to stress that here I am looking at the action of capital H, and
I am talking about H-orbit, so the point is capital H is the
stabilizer of A, so if small a is an capital A the H-orbit of small a
is completely inside A. So in other words,
(Refer Slide Time: 19:10)

hence, A is the union of H-orbits right, because remember H-

orbits are of course H-orbits are disjoint. And a must be the
union of, why is this, why is it a union of H-orbits? Because if a
belongs to A then the H-orbits of a, let us denote by H-orbit is
completely contained inside A. So what is H-orbit? I am going
to write it as ha, because it is small h times small a, as small h
varies in capital H, this is simply the right coset. Right, this is
right coset of a.

So H-orbit, if small a is a capital A the entire right coset is in

412
capital A. So A must be a union of Ha, of course Ha will be
same for different a’s perhaps, Ha maybe equal to Hb, but they
cover A, there is no problem. Now also we know that cosets are
disjoint, so we can write A as, in fact, we can write A as a
disjoint union of cosets.

So, right, we can write A, as some Ha1, disjoint union Ha2,

disjoint union H(ar), so because again remember cosets which
are equivalence classes for a group action are either same or
they are disjoint. So A is union of Ha, because every small a is a
some Ha, namely it is in Ha and they cover A of course.

Hence they cover A, the important point is Ha is, all of Ha is in

capital A, so we have this. And once you have that, we just
identify the distinct ones among them and write it like this. So A
is a union of Ha1, Ha2, H(ar), so in particular the order of A is
equal to order of Ha1 plus the order of Ha2, and so on up to the
order of H(ar). Okay, now let’s see, what is the order of Ha1?
Order of Ha1 is just order of H. Similarly order of Ha2 is order
of H, order of H(ar) is order of H again. So this is the order of H
times r.
(Refer Slide Time: 22:02)

So order of A is equal to r times order of H. This means exactly

that H divides order of H divides order of A, right. This is the
meaning of dividing and here we are only using the fact that, so
we used if you have Ha remember if you take a coset of a

413
subgroup, left or right it doesn’t matter, the order is the same
because it is a group property.
Small ha as you vary h, will be, there will be as many elements
as there are elements of capital H. So this is an easy exercise that
we have done in the previous videos, so this prove the
proportion right, so let’s recall the proportion or the lemma I
called it. If you have G acting on itself by left multiplication and
you take a subset of the group G and its, the subset is called as A
and H is the stabilizer of A , then the order of H divides the
order of A.

Okay, this is important for us when we do proofs of Sylow

theorems later on. So now let us go ahead and start our
discussion of Sylow theorems. So now our goal is to study.
(Refer Slide Time: 23:31)

So we are going to prove Sylow theorems. Okay, there are three

Sylow theorems. I am going to state them and prove them one
by one, but they all have to do with the following setup.

So G let G be a finite group. So I am going to, Sylow theorems

are very important theorems to understand something about the
finite groups. So we understand this in the following way.

So let p be a prime number, so we are going to write suppose the

order of G is n, so we are going to write n as p^e times m, where
p and m are co-prime. Remember that “co-prime” means p and
m have no common factors because p is a prime number it
simply means that p does not divide m, so this only means, in

414
this situation, p does not divide m.

Okay, we can always write any number like this. So n equals

p^e, power p power e times m. See for example if n is 6 and p is
2, we can write 6 as 2 times 3, if p is 3 and n is equal to 6, we
write 3 as 2, 3 times 2. But if n equal to 6, so here e equal to 1
and m equal to 3. So write 21 here also e equal to 1 but m equal
to 2. Now on the other hand if n is equal to 6 and p is equal to 5,
how do you write this? It is 50 times 6, so e is zero, m is 6.

So e is, in this description, e is always non-negative. Right, and

m is always at least 1, so n is at least 1. So its a finite group, so it
must have at least 1 element. So we have this situation, so we
simply look at how many times p appears in n, it may not appear
at all like when it is 5 and n is equal to 6, or if n is 12 and p is 2,
we have 12 equals 2 squared times 2, so e is 2 and m is 3. So we
simply look at how many times a prime appears.

So now Sylow theorems are really statements about special

subgroups of a group G and these subgroups are described by
the element p.
(Refer Slide Time: 26:35)

So the important definition for us in this is: A, so G is a finite

group as above, a “Sylow p-subgroup” of G is a subgroup of
order p power e. So we of course write, it is important to write

415
this, so the order of group is written as p em. So a Sylow-p
subgroup of G is a subgroup order p^e, where p^e is the largest
power of p that divides the order of the group G, okay. So a
Sylow-p subgroup is a subgroup of order p power e.

So as an example, a Sylow-2 subgroup of S3 is a subgroup of

order two. And if you recall our study of S3 in the previous
videos, S3 has 3 Sylow-2 subgroups, because it has three
elements of order two, each of them generates a subgroup order
2, so there are three Sylow-2 subgroups of S3. On the other
hand, S3 has only one Sylow-3 subgroup, so I want you to get
used to these notations, what is a Sylow-3 subgroup of S3? What
is the order of Sylow-3 subgroup? Remember 6 is written as 3 1
times 2, so a Sylow-3 subgroup has order 3. And there is only
one subgroup of S3 which has order 3.
(Refer Slide Time: 29:26)

And S3 has no Sylow p-subgroups if p is not equal to 2 and p is

not equal to 3. Because if p is not equal to 2 and p is not equal to
3, that prime does not divide the order of S3, which is 6. So p^e
will be p^0, so there is no subgroup of that order, so we only
consider this, so this is the convention. We only look at Sylow-p
subgroups, if p divides e, p divides order of group, in other
words in my notation small e is positive.
On the other hand if G is a group of order 100, a Sylow-5
subgroup of G, let us think about it, what is, see 5 is a prime
divide 100, so it Sylow 5-subgroups can be studied.

So what is a Sylow 5-subgroup of G? What will its order be? It

416
will be order 25. This is because 100 are equal to 5 2 times 4, 25
times 4. So a Sylow 5-subgroup will have order 5 squared.

Similarly a Sylow 2-subgroup of G has an order, now 4, because

100 is 2 squared times 25. So Sylow 2-subgroup will have order
4, 2 squared. So we look at the largest power that divides the
order, not just, for Sylow 2-subgroup it’s not 2, its 4. Similarly
Sylow 5-subgroup it is 25.

So these are Sylow p-subgroups of a finite group G and the

Sylow theorems are statements about their existence, their
properties, how many there are and so on. So I will stop this
video here, in the remaining videos we are going to state Sylow
theorems and prove them and then look at some examples and
exercises. Thank you.

Online Editing and Post Production

Karthik
Ravichandran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh Kumar
Web Studio Team
Anitha
Bharathi
Catherine
Clifford

417
 Deepthi
Dhivya
Divya
Gayathri
Gokulsekar
Halid
Hemavathy
Jagadeeshwaran
Jayanthi
Kamala
Lakshimipriya
Libin
Madhu
Maria Neeta
Mohana
Mohana Sundari
Muralikrishanan
Nivetha
Parkavi
Poonkuzhale
Poornika
Prem Kumar
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirely
Sorna
Subash
Suriyaprakash

418
 Vinothini
Executive Producer
Kannan Krishnamurty
NPTEL Coordinators
Prof. Andrew Thangaraj
Prof. Prathap Haridoss
IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resource Development
Government of India
HYPERLINK "http://WWW.nptel.ac.in" WWW.nptel.ac.in

419
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 08
Lecture 41 – “Sylow Theorem 1”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

So let us now continue our study of Sylow theorems, so with

whatever background I gave in the last video, let me now start,
go ahead and start with the first Sylow theorem. So there are
three Sylow theorems.
(Refer Slide Time: 00:32)

First Sylow theorem is the following. So let G be a finite group,

let p be a prime number such that, let p be a prime number such
that p divides the order of G. So we write the order of G, so we
write the order of G, let us call it small n, as p^e times m, where
because I am now assuming that p divides the order of G, e is
positive and p does not divide m, okay.

So remember that this means that we have taken out the largest
power of p that is available in the order of G.

So then this is the set up, as in the end of the last video, I set it
up like this. Then so actually this data is not relevant for the
statement, so the statement is, let G be a finite group, let p be a
prime number that divides the order of the group, then G has a

420
Sylow p-subgroup. So then G has a Sylow p-subgroup. So this is
a very strong theorem in the, you should recall first of all, recall
Cauchy’s theorem.

So if you recall, one of the theorem that I did in previous weeks

was Cauchy’s theorem. So there I assumed that if G is an
abelian group, finite always as always, and a prime p divides
order of G then G has an element of order p. So this was the
Cauchy’s theorem. So the first Sylow theorem is a vast
generalization of this thing because first of all I am not assuming
G it is an abelian group and in fact we are saying much more
than that it has an element of order p.

Because we are saying that in fact it has a subgroup of order p

power e. As a corollary after proving this, I will prove that first
Sylow theorem will definitely imply that any group of order, any
group whose ordered is divisible by p has an element of order p.
(Refer Slide Time: 03:41)

(Refer Slide Time: 03:42)

So first Sylow theorem is a, it is not, I mean it is a vast

generalization, so I mean it generalizes Cauchy’s theorem to
arbitrary finite groups and makes a stronger statement than
saying that there is an element of order p. So my goal today is to

421
prove that G has a Sylow p-subgroup, remember Sylow p-
subgroup is a subgroup of order p power e. So the proof of
Sylow theorem, so rest of the video will be focused on the proof
of Sylow theorem.

And as I mentioned in the previous video, this proof as well as

the proofs of next two Sylow theorems, depend critically on
action of G on itself and its subsets. The two actions we consider
are left multiplication and conjugation. So now I am going to
consider the following set, so earlier I talked about G acting on
the power set of G, G acting on the set of all sets of subsets of G.
Now I am not interested in all subsets.

I am interested in only subsets of cardinality p power e, so let

capital S be the set of subsets of G of order p power e. So
remember background, always we assume that n can be written
as, n is the order of group, and that is written as p^e power m, p
power e times m and p does not divide m, this is our set up. So
let us look at the subsets of order p power e, so in other words S
is all sets A and I again stress these are subsets only, not
subgroups.

The cardinality of A is p power e. G acts on S by left

multiplication. So remember in the previous video, I said g acts
on subsets of G by left multiplication, there I defined G (DOT)
A to be g times small a, as small a varies over capital A. Now I
am introducing a further restriction here, I am not looking at all
subsets but I am looking at subsets of order p power e and all
you need to verify here, to verify the G acts on S, is that, it is an
easy exercise actually to verify this. If A has p power e elements

422
then gA has also p power e elements. So in fact I should state
this exercise better.
(Refer Slide Time: 06:45)

So in fact I should state: order of A is order of gA. So if order of

A is p power e then order of gA is also p power e, and this is
repeatedly used in all the videos that we have done, because this
is related to the statement that two cosets of a subgroup have the
same number of elements, but here even if A is not a subgroup
the statement holds, because the set of elements ga will be
distinct if a is distinct.

So in other words if a and bare distinct ga and gbare distinct. So

the number of elements of A is equal to number of elements of
gA. So in other words if you take a subset of order p power e,
apply a group element to it you get another element of order p
power e, so G acts on this set S. So I am not interested in the
power set of A, I am interested in the subsets which have p
power e elements.

So now I am going to recall for you a fact, what is the order of

S? So this is a simple combinatorial argument that you may have
seen earlier or you can think about it and convince yourself, so I
am not going to prove this because this takes me on a tangent.
So you have n elements, small n many elements in capital G,
and you want to construct a group of, a subset of order p power
e, so the number of ways for doing this is simply n chose (p
power e).

423
So this is n chose (p power e), that you have studied I am sure in
other courses. So the number of elements of capital S is n chose
(p power e), in other words number of subsets of group G which
have cardinality p power e, precisely n chose (p power e), and
what is n chose (p power e)? It is n factorial divided by (p power
e) factorial times n – (p power e) factorial. So this if you cancel
out n – (p power e) factorial what you will have is n times n -1
times n-2 upto n-pe+1.

Because n – pe factorial will be cancelled from this and you will

have p power e factorial, so (p power e), (p power e)-1, (p power
e)-2 and 1 the last term will be 1. Which is, of course (p power
e) – (p power e)+1. So this is the number of elements of capital
S. In fact, this is the first fact. The second fact is.
(Refer Slide Time: 09:34)

Fact two is that p does not divide the cardinality of S, so this is

also not difficult at all but and I will give you a very quick
argument about why this is true. So remember that order of S is
this, right order of S is this. So there are in the numerator and
denominator, there are same number of elements. So think of
order of S as n by (p power e), n-1 by (p power e)-1, n-2 by (p
power e)-2. n- (p power e)+1 by (p power e)-(p power e)+1. So I
am just, because there are exactly same number of factors in
both numerator and denominator, I am going to write it like this.

So if I show that p does not divide each of them then we are

424
done but here of course why does p not divide? I mean that these
are not necessarily integers, but I claim that the same factor of p
divides n-i and (p power e)- i for any i from 0 to (p power e)-1.
Because that is the last factor, so each factor can be that of as n-i
where its n-0 here n-1 here, n-2 here, n- in bracket, so this can
be written as, so there are (p power e)-1, actually (p power e)
factors here n-0, n-1 and n-2,…

So same the factor divides, so if p divides n three times I claim

that it also divides p power e three times and then same thing
happens everywhere. So there cannot be after you cancel all the
things and all the factors and compute cardinality of S, there
cannot be a p in its order and this is easy to see because suppose
i can be written as p power i, sorry, let us say p power r
times I don’t know so k.
So let us say write i, as so here of course t could be 0 also. So
then n-i is (p power e)m, remember n is (p power e)m - (p power
r)k. So you can factor out and remember r must be strictly less
than e because i is strictly less than p power e, so r is strictly less
than e. So I can factor (p power r) and what I will have is (p
power e-r)-k, so r is the largest power that divides, of p, that
divides n-i. Similarly (p power e)-i will be (p power e)- (p power
r) times k. Again we have (p power r)(p power e-r)-1, okay.
So then if you write it like this, then (p power r) is the largest
power of p that divides both n-i and (p power e)-i. So this forces
capital S not to have, the order of S not to have any factor of p.
Because in each of these ratios the same p appears, so when
does p divide order of S? It divides it when one of this factors
has a p remaining, after you cancel p in both numerator and
denominator it should have some p left but that does not happen
because if p power 2 divides n-1 and p power 3 does not divide

425
it exactly same happens (p power e)-1 so then that is all. So it
turns of that, there is no factor of p in S, so I got to use this.
(Refer Slide Time: 14:23)

(Refer Slide Time: 14:24)

So the key fact for us, to be used later is, p does not divide, so
just if you understand this that is great, but I want you to now
take, spend a minute thinking about this argument why p does
not divide order of S. But please remember that if you don’t
understand, it will not affect the rest of the proof. So if you don’t
understand don’t get worried about it forget it for the moment,
accept this is a fact and try to follow the rest of the proof.

And if you don’t understand why this statement is true. You can
go back and read the proof, listen to the proof again carefully or
ask questions. So in the rest of the proof I am not going to use
any of these calculation, I am only going to use this fact: p does
not divide order of S. So let us only use this. So let us accept
this, so now we are ready to prove Sylow first theorem.

So what we have is: G acts on S by left multiplication. And what

is S? S is the set of all subsets of capital G which have order p
power e. So it acts on S. So now we have the orbit
decomposition of S. What does it say?

Order of S is order of orbits order of the first orbit some, I don’t

know, it doesn’t matter, so this is some k orbits. So where O1

426
through Ok are distinct orbits for the action of G on S. So any
time you have a finite group G acting on a finite set S we have
the orbit decomposition in hence order of S is the sum of the
orders of individual orbits.
But now by the above fact, p, which is the fixed prime number
we are dealing with, p does not divide order of S. Now look
closely at this equation: order of S is equal to order of O1+ order
of O2+ ... (dot, dot, and dot) + order of Ok; p does not divide
the left hand side, so p cannot divide all the terms on the right
hand side, right. Because if p divides order of O1 and order of
O3 and so on up to order of Ok, p will divide everything in the
right hand side, so p will divide the sum also. But then that
violates the fact that p does not divide order of S.
(Refer Slide Time: 17:45)

(Refer Slide Time: 17:46)

So there exits an order orbit Oi such that p does not divide the
order of Oi, right. This is clear because if p divides all orders p
divides the sum. Then p divides order of S which cannot happen,
so there must in other words the statement that p divides all
orders is wrong. Then that means p does not divide some order,
so p does not divide orbit Oi. So say remember, what is Oi?
These are orbits of elements of S. So let us say it is orbit of A, so
A is in S, so p does not divide orbit of A, that is the conclusion
for us.

427
P does not divide orbit of A. Now let H be the stabilizer of A
and our claim now is H is the desired subgroup. We claim that H
is a Sylow p-subgroup of G. Note that being stabilizer of A or
some element of capital S, remember A is an element of capital
S, capital S is a set of subsets, so an element of capital S is
actually a subset of G. So A is an element of capital S, H is a
stabilizer so H is definitely a subgroup.

The thing to prove here is we know that H is a subgroup of G.

We need to show that it is a Sylow p-subgroup. What is the
meaning of Sylow p-subgroup? We want to show that its order
is equal to p power e. In the last video I defined Sylow p-
subgroups for a group G and in our context it is exactly
subgroup of order p power e. So in order to claim prove the
claim which is that H is a Sylow p-subgroup we need to only
show that order of H is p power e.

So now I am going to use another small lemma that I proved in

the previous video. So what we know is the following. By the
lemma in the previous video, by the lemma in the previous
video, order of H divides order of A. So if you go back and look
at the previous video this is exactly the lemma that I proved, if G
is a group acting on subsets of G and you take a subset and you
look at its stabilizer, order of the stabilizer divides the order of
the set.
(Refer Slide Time: 20:52)

428
(Refer Slide Time: 20:53)

So H divides and what is the order of A? Order of A is p power

e because A belongs to remember A belongs to capital S which
is the set of all subsets of p power e, so order of H divides order
of P power e, sorry order of H divides p power e, so order or H
is a power of p, so if you have a number dividing p power e, p is
a prime number so it must be a power of p. So only number that
divide p power e are powers of p.

So this is because p is prime. Of course this is not true if p is not

a prime so if some number divides p power e it must be a power
of p by itself. We also have the counting formula. What is the
counting formula? It says that order of G, remember G acts on S,
use counting formula for the G action on S and the element A in
S. I recalled earlier in the video the counting formula it says the
order of the group is product of order of the stabilizer of A and
the orbit of A.

And what is the order of group? It is p power e times m, order of

the stabilizer which I called H is some power of, let’s say p
power i, let us say p power i is the order of H. So remember I
concluded that order of H is a power of p, so it is one p power i
and whatever is orbit of, order of the orbit. So and it is some
number so let us call it so may be number m not m so let us call
it r.

429
So p power e times m is equal to p power i time r. But remember
the assumption on orbit of A, orbit of A is such that p does not
divide the size of orbit of A, so by choice of A p does divide the
order of orbit of A, so in other words p does not divide r. So
now let us look at this carefully. So we have p^e m is equal to
p^i r, and what do we know? p does not divide r, so remember
integers can be factored uniquely. So you have p appears e times
on the left hand side.
(Refer Slide Time: 23:58)

So p appears e times on the left, so it must appear e times on the

right also. It appears i times her and it does not appear in r.
When I right “appear” what I mean is when I factor r into
product of primes p is not one of them. So this implies these two
facts imply that i equal to e, so because the p must appear e
times on the right hand side also and r cannot have any p’s in it.
So i must be equal to e.
So the order of the stabilizer is p power e, so we are done with
the proof the first Sylow theorem.

Recall the first Sylow theorem says if you have a group G finite
group G and a prime number p divides it, p must, G must have a
Sylow p-subgroup and we have produced it. Because we have
produced it because H is a Sylow p-subgroup of G, so this
completes the proof. The proof is very clever and you would not
normally think of proving it like this.

So what happened is we have looked at the set of all subsets of

430
G containing p power e elements. Of course some of them will
be subgroups but we didn’t directly prove that one of them is a
subgroup, what we showed is that stabilizer of one of them will
have order p power e. That is what we have shown. So there is
an orbit whose order is not divisible by p and stabilizer of that
element must have order p power e. So this proves the first
Sylow theorem and an important corollary of this is the
following.
(Refer Slide Time: 26:13)

(Refer Slide Time: 26:14)

So let G be a finite group and let p be a prime number that

divides order of G then G has an element of order p. So that is
my statement. This is exactly the generalization of Cauchy’s
theorem that I promised when I talked about Cauchy’s theorem.
Here I am removing an important word here I am not assuming
that G is an abelian group. We have already proved that if G is a
finite abelian group and a prime divides that order of that group
then that group has an order p element.

Now I am not assuming that the group is abelian anymore ,any

finite group has this property. Why is this?
By first Sylow theorem, G has a Sylow p-subgroup say H. So G
has a Sylow p-subgroup, say H. So in other words order of H is
equal to p power e and we write n as p power e m always p does
not divide m, n is the order of G. So now let us chose any

431
element of H what can be the order of a?

By Lagrange’s theorem the order of the element divides p power

e, order of the element divides the order of H which is, because
a is in H, Lagrange’s theorem. Lagrange’s theorem implies order
of a divides order of H which is p power e. So this in particular
means order of a is equal to p power r, for some, I am going to
assume that a is not identity, so r is between 1 and e. So it can’t
be 0, because if r is 0, p power 0 is 1, order of a is 1 means a is
e. So A is p power r, order of A is p power r. Now how do I
construct an order p element?
(Refer Slide Time: 29:00)

(Refer Slide Time: 29:01)

So now consider b is equal to a power (p power r-1), so I am

taking b to be a power p power r-1 which is of course an element
of H which is an element of which is a subset of subgroup of G.
So I claim that then we claim order of b is p. This is easy
because what is b power p? b power p is a power p power r-1
power p, and this obviously implies, this is equal to a power p
power r, because order of a is p power r this is e. So b power p is
e, right.
So now that doesn’t immediately prove that order of b is p,
because b power p is e means so order of b divides p. This is
something we have learned way back in the beginning videos. If
an element has certain power of an element is identity then order

432
of that element must divide that power, but only numbers that
divide p, p being a prime number are 1 and p, so order of b is 1
or order of b is p. If order of b is p we are done.
Can order of b is 1, can order of b be 1? Order of b 1 means
what? Only element of order 1 in any group is the identity
element. That means b must be equal to e, that means a power p
power r-1 is equal to e, that means order of a divides p power (r-
1) but order of a is p power r. That is by assumption, right, a was
an element of order p power r. Hence p power r divides p power
r-1, this is absurd. Obviously p power r cannot divide p power r-
1, so order of b is, p power, p and b is the element we are
looking for, so this proves the corollary.

So as I said this corollary generalizes the Cauchy’s theorem and

says that for any finite group, no longer needed we are assuming
it to be abelian, if a prime number divides the order of the group
then that group has an element of order p. And this is an
immediate corollary of the first Sylow theorem, so Sylow
theorem is way more than just saying that corollary. So it is
saying something stronger because it says that G has a subgroup
of order p power e.
So this hopefully gives you an idea of the power of Cauchy’s
theorem and the next two Cauchy’s theorems say further about
this, so first, sorry this says something about the power of Sylow
theorem and the next two Sylow theorems say further things
about Sylow p-subgroups. We know now that there is always at
least one Sylow p-subgroup, so in the next two Sylow theorems
we will study more properties of Sylow p-subgroups, Thank
you, I will stop the video here.

433
Online Editing and Post Production
Karthik
Ravichandran
Mohanarangan
Sribaliji
Komathi
Vignesh
Mahesh Kumar
WebStudio Team
Anitha
Bharathi
Catherine
Clifford
Deepthi
Dhivya
Divya
Gayathri
Gokulsekar
Halid
Hemavathy
Jagadeeshwaran
Jayanthi
Kamala
Lakshmipriya
Libin
Madhu
Maria Neeta
Mohana
Mohana Sundari
Muralikrishnan
Nivetha

434
 Parkavi
Poonkuzhale
Poorvika
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirley
Sona
Subash
Suryaprakash
Vinothini
Executive Producer
Kannan Krishnamurty
NPTEL Co-ordinators
Prof. Andrew Thangaraj
Prof. Prathap Haridoss
IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resource Development
Government of India
WW.nptel.ac.in
Copyrights Reserved

435
 NPTL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 08
Lecture 42-“Sylow Theorem 11”
Prof KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

So let us now, in this video, do the second Sylow theorem.

(Refer Slide Time: 00:21)

We have done the first Sylow theorem last video, so second

Sylow theorem. The first Sylow theorem remember said that if
you have a group, a finite group G and prime number p dividing
the order of the group then G has a Sylow p-subgroup. Second
Sylow theorem says that, let G be a finite group and let, this is
the same assumption always, let p be a prime that divides order
of G, then any two Sylow p-subgroups are conjugate, okay.

So let me, maybe I mentioned this usage earlier, but we say that,
I am going to say it again, we say that two subgroups H and K
are conjugate, if H equals some gKg inverse, for some g in G,
okay. So conjugation is this operation right, g sometimes,
something times g inverse. So we say the two subgroups are
conjugate if gK g inverses is H.

436
So the Sylow theorem says that any two Sylow subgroups are
conjugate, so all the Sylow subgroups in other words are
conjugate, because being conjugate is an equivalence relation. If
H1 and H2 are conjugate, H2 and H3 are conjugate, H1 and H3
are also conjugate. And as we will see later in applications, this
is a very important statement that, Sylow subgroups, Sylow p-
subgroups are always conjugate.

So let us go ahead and take two Sylow subgroups, let H and K

be two Sylow p-subgroups of G, remember Sylow I guarantees
the existences of a Sylow p-subgroup, but there can be more
right, we are not saying there is exactly one, so let us take two of
them, our goal is to show that they are conjugate, so now I am
going to consider the following.
(Refer slide Time: 02:58)

So we will consider, the actions of, let’s say, K on G/H by

conjugation. So as I have been saying repeatedly all Sylow
theorems and their proofs are essentially done by playing with
various group actions, different groups, different sets, different
actions.

So here my goal is to, my focus will be on the action of K on

G/H by conjugation. What is G/H by the way, these are all
cosets of H, left coset by our conversion, so these are gH where
g is in G.

So how do we define the action, so let say b is in K, so I am

going to use a for consistence, a as an element of capital G and

437
aH is an element of G/H, remember that in order to define the
action of a group K on a set G/H, I need to tell you what is a
group element times the set element. So b times aH, no surprise
here, it is simple ba times H, okay b times aH is define to be ba
times H, and easy to check that this an action. This is an action
of K on G/H, because identity element acts as identity and the
associativity naturally holds okay .
(Refer Slide Time; 04:46)

So what we have is counting formula. Says that, this is not the

counting formula actually, this is the remember G/H is a union
of disjoint K-orbits, I am going to use the notations K-orbits to
denote that it is actually action of K, usually we talk about the
action of G, so here I am looking at action of K, so I am going to
stress that by talking about K-orbits, so G/H is a union of
disjoint K-orbits, so that means we can write the order of G/H as
order of O1, order of O2, order of Ok. Let me use some other
letter here let say Or, here O1,…, Or are the distinct K-orbits of
G/H, okay.
(Refer Slide Time: 06:04)

Now let’s see, what is G/H, order of G/H, so remember always

we will use this notation, order of G is pem and what is the order
of H? H is a Sylow p-subgroup right, so order of H is p power e,
this implies order of G/H is m. This is our original counting
formula, so this is of course not divisible by p, that was how we

438
write this. p does not divide m okay. So now in other words, if
you look at this equation, this sort of thing happened exactly as
it is, exactly like this, in the first Sylow theorem. We have G/H
order is a sum of some numbers, p does not divide the order of
G/H, so p cannot divide one of the orbits sizes.

So p does not divide the order of some OI, for some I, because if
p divides order of each OI, then p divides the sum which means
p divides order of G/H, which is not possible. So p does not
divide OI, order of OI, for some I. So say OI is the orbit,
remember what is OI is, it is convenient to write it like O1, O2,
O3, but they all orbits of elements of G/H. So say OI is the orbit
of sum aH. Remember the set in question here is G/H, so orbits
are element, orbits of elements of G/H. So say OI is the orbit of
aH, so in other words, what we have is that, p does not the
divide the orbit of aH.

So now let’s apply the counting formula. Now the counting

formula applied to the K-action on G/H, what does the counting
formula and this particular element, counting formula remember
is applied to a specific element of the set, which I am taking to
be aH.
(Refer Slide Time: 08:31)

So it says that the order of K is equal to order of stabilizer of aH

times order of orbit of aH. Again what is the order of K? Order
of K is p power e, that is because K is a Sylow p-subgroup so it
must have order pe. So this is equal to order of stabilizer of aH

439
times orbit of aH. But remember what we have here, p does not
divide the order of orbit of aH, but p does not divide the orbit,
the order of orbit of aH, because that is how we chose this, p
does not divide order of some orbit and we called it orbit of aH.

So now let us look at this equation: p power A is equal to the

product of stabilizer order of stabilizer of aH and order of orbit
of aH, but p does not divide orbit of aH.
(Refer Slide Time: 09:44)

So stabilizer of aH must have order p power e, so again the

uniqueness of factorization of integers, p power e is equal to
some number times another number, the second number cannot
be divisible by p and it of course cannot have any other prime
factors because p is only prime factor, on the left hand side, so
that means all of p power e must be present in the order of
stabilizer of aH, so order of orbit of aH is 1, order of stabilizer of
aH is pe.

But remember stabilizer of aH, where is it living? It is living in

K because we are looking at K-actions. Stabilizers are subgroups
of K but K has already order pe, stabilizer also has order pe, so
stabilizer of aH is exactly equal to K right. The entire group is
the stabilizer of aH, but what does this mean, this means that b
times aH.

So we have the following, b times aH is equal to aH for all b in

K. b times aH is equal to aH for all b in K. This is the exactly
the meaning of every element of K being the, being in the

440
stabilizer of aH. This means (ba)H is equal to aH, for all b in K
right. But this means (ba) H contain ba so ba belongs to aH, for
all b in K right.
(Refer Slide Time: 11:38)

So I am just going step by step here, hopefully each step is clear,

so ba is in aH. That means, b is in aH a inverse for all b in K,
right. Because small b times small a is equal to, small a times
small h, I can now multiply by a inverse on the right for both
sides to get b is equal to a small h, a inverse. So b is in aH a
inverse. This is true for every b in K, that means K is a subset of
aH a inverse right, because every small b in capital K is in aH a
inverse. So all of K is in aH a inverse. But this must mean that K
must have the same order K, remember what is the order of,
okay.

So let me right it like this. Note, order of K is p e, order of H is pe

because both are Sylow p-subgroups, but order of H is pe order
of aH a inverse is also p power e, because conjugation also does
not change the order. This is a simple exercise for you, but K is
inside aH a inverse, but both have the same number of elements.
So K is equal to aH a inverse. Remember this is the exactly what
we wanted to show.
(Refer Slide Time: 13:11)

What did we want to show, any two Sylow p-subgroups are

conjugate. So this completes the proof of the second Sylow

441
theorem.

So in other words, hence the second Sylow theorem shows that

all Sylow p-subgroups are conjugate to each other. So you take
any two Sylow p-groups, apply the theorem to show that they
are conjugate to each other.

This is a very strong statement and I will just give you quickly
two remarks here. So first of all, as an example let us take S3
and I mentioned in an earlier video that G has 3 Sylow -2
subgroups okay.
(Refer Slide Time: 14:30)

So if you use the cycle notation for S3, remember it is { e (12)

,(13), (23), (123), (132)} and the Sylow 2-subgroups are {
e,(12)} , H2 will be {e,(13)} , and H3 is {e,(23) }. These are the
Sylow 2-subgroups because remember Sylow 2-subgroups must
have order 2 because 6 is 2 times 3.
(Refer Slide Time: 15:09)

So now the Sylow, second Sylow theorem says that H1, H2, H3
are all conjugate to each other, okay. This I will leave as an
exercise for you to specifically choose an element which
conjugates H1 to give you H2, okay. As an exercise, may be I
will just ask this, find an element, let’s say a in S3 such that a
H1 a inverse is H2. By the Sylow’s theorem we know that the
H1, H2 are conjugate. In this example I want to explicitly find

442
such an a.
(Refer Slide Time: 16:01)

And an important corollary I will now mention is the following.

Suppose if a group G, a finite group of course always, has only
one Sylow p-subgroup, H let’s say. So G is a finite group and H
is the only Sylow p-subgroup of G. Then H is normal G, okay.
(Refer Slide Time: 16:37)

Why is this? So the proof is the following. Recall what is a

normal subgroup? To prove H is normal in G, we must show,
what do we need to show, we need to show that if g belongs to
G, h belongs to H then gfg inverse is in H. Equivalently we have
to show, show that g capital H g inverse is equal to H, for all g
in G. That is what one has to show to prove some subgroup is
normal.

Now if H is a Sylow p-subgroup then H will have order p power

e, of course again order of G contain p power e as the largest
power of e. So now if g is, small g is in capital G then as I have
said before the order gH g inverse is p power e and it is easy to
show that gH g inverse is a subgroup of G. See if H is a
subgroup gH g inverses is a subgroup. See the cosets are not
subgroups but gH g inverse, conjugate are always subgroups
because identity is there. Remember g times e, small e is in H so
g times e times g inverse is e and any two things like this are,
their product is also a thing like this, inverse is also a thing like

443
this.

So it is an easy exercise to show that gH g inverse is a subgroup

of G and it has p power e elements. So gH g inverse is a Sylow
p-subgroup, remember Sylow p-subgroup is a fancy term for just
a subgroup for order p power e, gH g inverses is a Sylow p-
subgroup because it is a subgroup and it has a order p power e.
But what was our assumption in the corollary, G has only one
Sylow p-subgroup.
(Refer Slide Time: 19:06)

By hypothesis, there is only one Sylow p-subgroup, namely H.

So gH g inverse is H and that proves that H is normal.

As an example, so if you somehow conclude that there is exactly

one Sylow p-subgroup, that Sylow p-subgroup must be normal.
If you take S3 and p to be 3, so 6 is 3 times 2, and we know that
and I can, I mean I mentioned this earlier. We know that there is
only one Sylow 3-subgroup, namely H will be {e, (123) and
(132)}.
(Refer Slide Time: 19:59)

Of course, in this case we can directly check that H is normal,

but it follows from the by corollary that H is normal. Note that
any conjugate of H is a subgroup of order 3, hence it is a Sylow
3-subgroup. But Sylow 3-subgroups there is only one such

444
thing, so H is normal. In this case it is also clear that H is normal
directly, but I wanted to give a simple exercise, example to show
that, and we will see more examples later, it is very useful know
that is there is one Sylow p-subgroup, in which case it will
automatically be normal, okay.

I will stop the video here. In this video we talked about Sylow,
second Sylow theorem, which said that any two Sylow
subgroups are conjugate. In the next video I will talk about the
3rd Sylow theorem and which talks about the number of Sylow
subgroups. As a corollary here, as the corollary here shows, if
you know that there is only one Sylow p-subgroup we know it is
normal. So that is useful to know. So the third Sylow theorem
tells something about the number of Sylow p-subgroups. So I
will stop the video here, thank you.

Online Editing and Post Production

Karthik
Ravichandran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh Kumar
Web Studio Team
Anitha
Bharathi
Catherine
Clifford
Deepthi
Dhivya

445
 Divya
Gayathri
Gokulsekar
Halid
Hemavathy
Jagadeeshwaran
Jayanthi
Kamala
Lakshimipriya
Libin
Madhu
Maria Neeta
Mohana
Mohana Sundari
Muralikrishanan
Nivetha
Parkavi
Poonkuzhale
Poornika
Prem Kumar
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirely
Sorna
Subash
Suriyaprakash
Vinothini
Executive Producer

446
 Kannan Krishnamurty
NPTEL Coordinators
Prof. Andrew Thangaraj
Prof. Prathap Haridoss
IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resource Development
Government of India
HYPERLINK "http://WWW.nptel.ac.in" WWW.nptel.ac.in

447
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 08
Lecture 43 - “Sylow Theorem 111”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

So, let us continue now. In the previous two videos we looked at

the first and the second Sylow theorems. The first one said that a
group, a finite group G has a Sylow p-subgroup, if p divides the
order of the group. Second Sylow theorem said that any two
Sylow p-subgroups are conjugate.
(Refer Slide Time: 00:32)

The third Sylow theorem says something about the number of

Sylow p-subgroups. So let us prove this, let us first state this.

Let G be a finite group. The set up is as in the first two Sylow

theorems. So, let G be a finite group and let p be a prime that
divides order of G. So we write order of G as pe m, where p does
not divide m okay, as always. Let small s be the number of
Sylow p-subgroups.
So, as I said the third Sylow theorem is a statement about the
number of Sylow p-subgroups. Let us call it s, remember s it at
least 1, by the first Sylow theorem. Then the third Sylow
theorem says two things. (1) s divides m. Okay, so remember m

448
is the factor of, in the order of the group, which does not contain
any p, after removing all the factors, largest power of p that is
available. So, s divides m and (2) s is of the form ap+1 for some
a in natural numbers.

Okay, so that means when you divide s by p, you have

remainder 1. So, let us go ahead and prove this. Proof is again a
clever use of group actions as the first two Sylow theorems
work. Here also we cleverly use a group action on a suitable
subset. So, here I am going to consider the set, the useful set
here is capital S is the set of all Sylow p-subgroups. So this is, H
is a subgroup of G and order of H is p e, that is the set S.
(Refer Slide Time: 03:14)

Now we want to consider a G-action on S. So, remember G acts

by two things on itself or subsets, either by left multiplication or
by conjugation. Here if you take left multiplication it won’t give
you an action, why? Remember S is not just subsets, S is not the
set of subsets. It is a set of groups; it is a set of subgroups. So, if
you take left multiplication, you are no longer going to be in the
set because if you multiply a subset, subgroup by an element on
the left you will have a subset not necessarily a subgroup.

So, if you consider the action of G on S by left multiplication it

will not give you an action because our set now consists of
subgroups. So, we are going to consider the action by
conjugation. Remember as I said in the previous video,
conjugation preserves this set. So, in other words if H is an S,
then g and g is in G, gH is also, gH by definition is gH g −1. So
here it is not the left multiplication, it is the group action, g. H is
also in S, because gHg−1 is a subgroup and its order is also pe.

449
So, it is an action. So, now this is the goal for us. Now by the
second Sylow theorem, so first of all let us fix some element of
H, S so. Let H be in S. So, that is H is a Sylow p-subgroup of G.

What is the orbit of S, sorry orbit of H? What is the orbit of H

under this action? Remember the proof in this case completely
depends on the action of the group G on this set S by
conjugation. What is the orbit of H? It is all gH g−1, as g varies is
in the group, gH g−1. Now remember gH g−1are all Sylow p-
subgroups that is okay. But by the second Sylow theorem this is
all of S, correct?

Because every, any two Sylow p-subgroups are conjugate. So

you take any element of S, namely a Sylow p-subgroup. It is
conjugate to H because any two conjugate; any two Sylow p-
subgroups are conjugate by second Sylow theorem. So, there
will be a suitable g such that gH g−1 is that Sylow p-subgroup.

So, the orbit of H is all of S. What is the stabilizer of H?

Stabilizer of H is all group elements such that gH g−1is H. Under
the action you must get back H. This is what we called the
normalizer, if you recall from one of the earlier videos, this is
the normalizer of H. So, let us denote this by N (H) and for
simplicity I am just going to call it N, okay.
(Refer Slide Time: 06:42)

450
So now the counting formula gives me the following. Counting
formula applied for, applied to this particular G action on S and
for the element H. It says that cardinality of G is equal to
cardinality of stabilizer of H times the orbit of H. This is the
counting formula. This is pe times m; this is equal to, what is the
stabilizer of H? This is the order of N, N was my notation for the
normalizer and orbit of H is S, right.

Orbit of H by the second Sylow theorem is all of S. So hence,

pem is equal to capital N times capital s. So, capital N order
capital N times remember our notation was number of Sylow p-
subgroups was called s, right in the definition. In the statement
of the theorem small s denotes the number of Sylow p-
subgroups. Capital S the set of Sylow p-subgroups. So, order of
S is small s; so pe m is s times order of N.

Now what we have is the following. Normalizer of a subgroup H

will always contain the subgroup. So we have this because
remember normalizer is all elements such that gH g−1= H, if
some element is already in capital H, certainly it will contain, it
will be contained in the normalizer so, we have this. Then I
claim, so we have the index of N in G divides the index of G in
H, H in G. Why is this?

I will quickly tell you why. What we have is the counting

formula for the subgroup of a group tells me that, this is one of
the original counting formulas. Order of the group is the index
times order of the subgroup. But remember this is also equal to
the index of N in G times order of N. I am applying the counting
formula first to H and then to N.

451
(Refer Slide Time: 09:22)

But since, H is in N, Lagrange’s theorem says what? Lagrange’s

theorem says that order of a subgroup divides order of the group.
So order of H divides order of N. So we can write order of H
times some, let us say L is order of N for some L in natural
numbers, positive integer.

So, now let us recall this, so order of G is index of H times order

of H, which is the index of N times order of N. But order of N is
order of, so I am going to rewrite this as index of N and instead
of N order of N, I am going to use this, order of H times L.

So, now I will look at the, this term and this term and cancel H,
order of H. So, index of H is equal to index of N times L. So,
index of N divides index of H. Okay, index of N divides index
of H. Now let us look at the counting formula that we had
earlier. What is the index of N?
(Refer Slide Time: 11:15)

So I want to understand, by the counting formula we have order

of G is equal to s times order of N, right. So this implies s is
equal to order of G divided by order of N, which is index of N.
But what I have just concluded is, index of N, so is that clear?
We proved earlier that order of G is equal to small s times order
of N. So, small s is order of G divided by order of N which is by

452
definition the index of N, not by definition but by the counting
formula, is index of N.

But we know that index of N divides index of H. And what is

the index of H? This in turn is the order of G divided by order of
H, but H is remember a Sylow p-subgroup. So, this is pem by pe
, this is m. Okay, so s is the index of N so, s divides m and I
think this gives (1), I called it (1), (1) of the theorem.

Remember the statement of the Sylow theorem is that if you call

small s, the number of Sylow p-subgroups it divides m and it is
of the form ap+1.

So, I have just proved the first part of it. Right, this gives the
first statement of the theorem.

To prove the second theorem, second part to prove (2) I am

going to consider a different action by a different group. To
prove (2) we consider a different action. So, here I want to
consider, we look at, so earlier I looked at the action of the
group G on the set of all Sylow p-subgroups that gave me the
statement (1). To prove statement (2) I will look at the action of
H, H is a, so we actually fix. So, fix a Sylow p-subgroup, call it
H, and we look at the action of H on the same set as above.
(Refer Slide Time: 13:58)

So, S being all Sylow p-subgroups of G, right. So as in the

previous, part (1) of this theorem, S was the same set, I am only

453
changing the set the group which is acting. I do not consider the
action of G, all of G, I only consider the action of the group H.
So, Sylow p-subgroups the set remains the same. And I am
going to call this set H, see H is an element of this right? So, H1
is called, I am just going to take H as H1, H2, H3 and remember
there are s of them. Small s is the number of elements of capital
S.

So, we look at H action on S, H action on S by conjugation. So,

only difference from the first part is, instead of looking at the
action of G on S by conjugation, we look at the action of H on S
by conjugation.

Okay, so now what is the, so I am going to figure out what is the

orbit of H? First I am going to understand what is the orbit of H.
What is orbit of H? So, orbit of H, so orb of H is all gH g −1, but
now where is g?

Because so maybe I should call what is the H-orbit of H.

Remember I am no longer interested in the action of G, what
happens outside H is of no concern for me. I am only interested
in the H-orbit of H. What is the H-orbit of H? It is all groups of
the form gH g−1, where g is in H. But this is exactly H. Right,
because if g is in H, gH g−1is H. So, orbit of H is just the
singleton H. Now choose some let any i be any number bigger
than 1.

What is the orbit of (Hi), is what I want to think about, and for
that I will suppose that {Hi} is an H-orbit of H, of S rather.
Remember I am looking at capital H acting on capital S by
conjugation. Suppose that some Sylow p-subgroup, different

454
from H, so (i) is greater than 1, different from H, let us say Hi
forms an orbit by itself. What can we say about Hi is what I
want to understand.

(Refer Slide Time: 17:04)

But if {Hi} is an orbit by itself, then what we have is h times

(Hi) times h−1 is equal to (Hi) for all h. Right, this is the meaning
of capital {Hi} being an H-orbit because H orbit of (Hi) is all
elements like this, all subgroups like this but it is just (Hi). The
singleton {Hi} is an H orbit, that means H times small h times
(Hi) times small h −1 is equal to (Hi) for every h in H. That
means small h is in normalizer of Hi for all (i). Right small h is
in normalizer of Hi for all i, because what is a normalizer?

Recall, I am going to call it Ni which is the normalizer of (Hi) is

by definition all g in G such that gHi g −1is Hi. If you go back
and see in the earlier part of the video, normalizer was defined
to be this. Now if h times (Hi) times h−1 is (Hi), then h belongs
to the normalizer. So, H for small, for every, sorry, it is for all
iH in capital H. That means capital H is in normalizer of (Hi).
So, capital H is in Ni. So, now we have two things.

So, we have H and (Hi) are both subgroups of Ni, right.

Because I concluded here that H is in Ni and I definitely know
that (Hi) is in Ni, that is not a problem. (Hi) is in Ni, H is also in
Ni, so they are both subgroups of Ni. Moreover H and (Hi) are
in fact Sylow p-subgroups of Ni. Why is that?

455
What are Sylow p-subgroups of Ni? You look at the order of Ni,
you look at the largest power of p that appears in it and you look
at those subgroups of that order. Remember Ni is a subgroup of
N, so order of Ni divides order of N.

So, the largest power of p that appears in Ni cannot be more than

e which is the, so in fact what we have is also some Ni is p em
prime, because remember what we have is Hi is contained in Ni,
contained in G. So, this has order, Hi as order p e, G as order p
e
m. So, order of Ni must be divisible by pi by Lagrange’s
theorem and in turn, it must divide P em. So, it must be some m
prime. Right so Sylow p-subgroups of Ni are also subgroups of
order p i.

H and Hi remember are subgroups of order p i. They are both

subgroups of Ni and they both have the order p i, so they are both
Sylow p-subgroups of Ni.

So, by the second Sylow theorem, so just to repeat, the largest

power of p that appears in Ni, is also p e. So, Sylow p-subgroups
of Ni are subgroups of order pi and we have shown we know
that H and (Hi) are both subgroups of N and both have order p
power; they are both subgroups of Ni and both have order p e.
So, they are both Sylow p-subgroups of Ni. Now what does the
Sylow second theorem say? It says that any two Sylow
subgroups or conjugate.
(Refer Slide Time: 21:30)

456
So, H and Hi are conjugate in Ni. So, here I am only focused on
applying the theorems to Ni, not to G. So, apply the second
Sylow theorem to Ni. Ni is a finite group; H and (Hi) are
subgroups of Ni which are both Sylow p-subgroups. So, they are
conjugate in Ni. That is, there exists g in Ni, such that gHi g−1 is
H, right. This is the meaning of being conjugate. But if g is in
Ni, recall what is Ni; Ni is the elements g such that gHi g−1 is
equal to Hi.

This means gH g−1, gHi g−1 is Hi but this is also at the same time
equal to H because gHi g−1 is H that is the assumption that, that
is the consequence of the second Sylow theorem because they
are conjugate but g being Ni forces gHi g−1 to be (Hi).This is the
definition of Ni. That means H is equal to Hi.

So, the upshot of all this is: if {Hi} is an orbit, is an H-orbit in S,

then Hi equal to H. Okay, so if Hi is an N orbit, Hi is an H-orbit
of S then it is H. what we have shown is H is an orbit by itself
that is okay but if Hi is an orbit by itself we must have that Hi is
equal to H. So, it is the only singleton orbit of S, so written
differently H is the only singleton orbit in S. Singleton orbit
means an orbit consisting of a single element. It is only singleton
orbit of S.
(Refer Slide Time: 23:54)

Now consider orbit decomposition of S. So, we have the size of

S which remember was denoted by small S is order of O1+ order
of O2+ …+ order of Ok. And let us assume that order of O1,
sorry O1 is orbit of H. H was the fixed thing that we had dealt

457
with. Okay, so that is H. So, order of O1 is 1 +order of O2 +
order of Ok. So, here of course I am looking at H-orbit
decomposition and by what we have done above by the above
argument, order of Oi is at least 2 for all i from 2 to k, right. So
every other orbit has at least two elements because if it has only
one element, it must be H. So, it has to, and that we have
already accounted for. So, they are all at least 2.

But now apply counting formula. What we get in the counting

formula? Here we are looking at the H action. So, let us say O i is
the orbit of Hi. So, if that applied counting formula applied to
that says order of H is stabilizer of O i, sorry stabilizer of Hi
times the size of Oi. But what is the order of H? H is a Sylow p-
subgroup, so order of H is pe.

So, this means order of Oi divides p, this is great, order of Oi

divides p because pe is stabilizer of Hi times order of Oi. So
order of Oi divides p. But for i at least two, we have two facts:
order of Oi is at least two by this argument and it also divides p.
Okay, so it cannot be 1, so they are all at least two. So, now let
us look at the orbit decomposition of S and I am going to write it
like this.
(Refer Slide Time: 26:41)

Small s is the order of S which is O1+O2 +Ok but we agreed that

order of O1 is 1 + order of O2 is something times p. Right,
something times so let us say order of O i is p times let us say Ai,

458
that Ai is at least 1. So, it is p times A2, p times A3, so order of
O3 is p times A3, order of Ok Ak is p times Ak. So, 1 times + p
times (A2 + A3+ Ak) and you call this A, and this is exactly the
statement that we made, this gives (2).

So, the proof is complete. Right, so we have shown that the

number of Sylow p-subgroups when you divide by p leaves a
remainder of 1. So, let us quickly recap this. The argument was
we look at; we fix a Sylow p-subgroup. So, let’s see where we
started with this. So, we fix a Sylow p-subgroup and then look at
the action of that Sylow p-subgroup on the collection of all
Sylow p-subgroups by conjugation and after some work we
concluded that the only singleton orbit is H itself. Any other
orbit must contain at least two elements.
So, the orbit decomposition looks like this: s equals 1+ the
remaining orbits. But the remaining orbits must divide p because
of the counting formula. So, all the remaining terms are divisible
by p and 1 is left so s itself must be of the form 1+ ap.

Okay, so let us again look at the example of S3 with that we

have seen before. So, here of course order of S3 is 6 which is 2
times 3. So, the number of, let us denote by s2, the number of
Sylow 2-groups, subgroups and s3 is the number of Sylow 3-
subgroups.
(Refer Slide Time: 29:26)

The third Sylow theorem says so, let us first do s2. So s2 divides
3 and s2 is 1+3A or 1+2A right. But, and what is s2? So, third
Sylow theorem says this. But we know what s2 is. What is s2?

459
We already seen right there are 3 Sylow 2-subgroups, that does
three satisfy this? Yes of course it does because 3 divides 3 and
3 is of the form 1+2 here.
What does third Sylow theorem says about s3? We know of
course we know s3 is 1 so, let us see if that is confirmed by
Sylow theorem, third Sylow theorem. s3 should be a divisor of 2
and s3 must also be 1+3a. So, 1 satisfies this. Correct, because 1
divides 2 and 1 is of the form 3A, 1+3a.

Okay, so this says that, this just confirms what Sylow theorem
says but I want to stress one fact here. Sylow theorem does not
tell you exactly what the number of Sylow p-subgroups is.
Okay, it only gives you a range of possibilities. You have to
further investigate the group to determine which possibility
occurs. See as in this example, before you knew this, suppose
you did not know this, we can say that even if we did not know
anything about s3, we can say s3 must be 1.

Why is that? Because s3 divides 2 that leave only 2 possibilities,

it is either 1 or 2. Right, but it also at the same time must be of
the form 1+3a but 2 are not of the form 1+3a. So, we can say s3
must be 1. This immediately come, no matter what the group is,
even if you did not know that is S3, s3 must be 2. On the other
hand s2 can be 1 or 3, why is that? Because remember the
conditions are s2 divides 3 and s2 is 1+ 2a.

Certainly 1 divides 1 has these properties because 1 divides 3

and 1 is of the form 1+2 times 0; and 3 also has this property, 3
divides 3 and 3 is of the form 1+ 2 times 1.

460
So, Sylow theorem only says that s2 can be either 1 or 3 and in
the example of the symmetric group, the option 3 is achieved.
Of course my notation is confusing here because the group and
also S3 so I hope you do not get too confused about this. So, if
the group is S3 then s2 is 3.

(Refer Slide Time: 33:07)

On the other hand let us take the group to be, if we take the
group to be Z mod 6Z, this is also group of order 6. Then s2 is
actually 1. This is an exercise for you; Z mod 6Z has exactly one
Sylow 2-subgroup. Okay, so in this case s2 is equal to 1 is
achieved.

So, just to recap Sylow theorem is not a conclusive answer to,

third Sylow theorem is not a conclusive answer to the question
of how many Sylow p-subgroups exist because it only gives you
a range.

As this is example of order 6 suggests sometimes that range

contains only one element like in this case; if you have a group
of order 6 we know that there must be only one Sylow 3-
subgroup that we can say because that number has to divide 2
and at the same time be of the form 1+3a.

On the other hand, when you look at Sylow 2-subgroups there

are either 1 or 3, and as you can see in these two examples, both
options are achieved.

461
So, the Sylow third, third Sylow theorem is not very precise, it
gives you a range of possibilities. Sometimes that range is very
small, in fact it can be even 1 possibility, sometimes you can
further study the group to eliminate some possibilities. But as it
is third Sylow theorem is very useful. So, I will stop the video
here in the next video I will look at some applications of Sylow
theorems and solve some problems. Thank you.

Online Editing and Post Production

Karthik
Ravichandran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh Kumar
Web Studio Team
Anitha
Bharathi
Catherine
Clifford
Deepthi
Dhivya
Divya
Gayathri
Gokulsekar
Halid
Hemavathy
Jagadeeshwaran

462
 Jayanthi
Kamala
Lakshimipriya
Libin
Madhu
Maria Neeta
Mohana
Mohana Sundari
Muralikrishanan
Nivetha
Parkavi
Poonkuzhale
Poornika
Prem Kumar
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirely
Sorna
Subash
Suriyaprakash
Vinothini
Executive Producer
Kannan Krishnamurty
NPTEL Coordinators
Prof. Andrew Thangaraj
Prof. Prathap Haridoss
IIT Madras Production
Funded by

463
 Department of Higher Education
Ministry of Human Resource Development
Government of India
HYPERLINK "http://WWW.nptel.ac.in" WWW.nptel.ac.in

464
 NPTEL
NPTEL ONLINE COURSE
Introduction to Abstract
Group Theory
Module 08
Lecture 44- “Problems 9”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

So, in the last few videos, we looked at Sylow theorems, proved

them, there are three Sylow theorems, remember that, they
describe Sylow p-subgoups of a finite group G. The first one
says there is always a Sylow p-subgroup, the second one says
any two Sylow p-subgroups are conjugate, and the third one
says the number of subgroups must satisfy, some conditions, so
there is a range of possibilities.

So in this video, I am going to do some problems and these

problems will illustrate how to apply Sylow theorems, so the
first thing that I want to do
(Refer Slide Time: 00:53)

let us called it a proposition. So I will show that any group of

order 15 is cyclic okay. So this is a very strong statement,
remember if you have a group of prime order, we already know
it has to be cyclic, because you take any element that is not
identity, its order must divide the prime number so it must be all
of the, it must be that prime number, so the subgroup generated

465
by that element must be the whole group, but certainly that
arguments fails for a group which has non-prime order like 15.
So how do we show that a group of order 15 is cyclic, so in
another words, we are saying that there is an element of order
15, and that is not clear.

All we know there is a element of order 5, and there is an

element of order 3, but how do we know there is an element of
order 5. So, and in this where Sylow theorems come in to play.
So we will show that if G is a group of order 5, 15 rather, then G
isomorphic to Z/3Z x Z/5Z, okay. So I do not remember whether
I have talked about the products carefully before so, quickly we
describe the product group.

It is very simple, so Z/3Z x Z/5Z, so remember that Z/3Z is a

cyclic group order of 3, Z/5Z is a cyclic group of order 5. So we
take elements (a, b), the first one in Z/3Z, the second one in
Z/5Z, and we add them, so I am going to use additive notation
here, so it does not matter, what notation you use, but we do
component-wise okay.

So, in other words (a, b) + (c, d) simply is (a+ c, b+d) okay. And
inverse of (a,b) will be (–a,-b), identity will be (0,0), so Z/3Z x
Z/5Z is a group under this. So I am going to say that any group
of order 15, this symbol stands for isomorphism, is isomorphic
to this. Remember that there is always a group of any given
order, namely a cyclic group of that order, so this statement
applies to that cyclic group also, so every group is isomorphic to

466
this, hence to the cyclic group.
(Refer Slide Time: 04:11)

So, now let us work with an arbitrary group, let G be a group of

order 15. So let us look at how 15 decomposes, as a prime
number, prime factoriazation, it is 3*5. So, G has at least one
Sylow 3-subgroup and Sylow 5-subgroup right.

Let us say as s3 is the number of, am going to use small s 3

number of Sylow 3-subgroups, and s5 is the number of Sylow 5-
subgroups. So now, the third Sylow theorem says what? The
third Sylow theorems says, s3 divides 5 and s3 is 1+3a, right. This
forces has s3 to be 1, remember if s divides 5, it must be either 1
or 5, but 5 does not have this property. So, s3 is 1.

Similarly s5 divides 3, s5 is 1+5a, remember a is some integer of

course that is why 5 cannot be return as 1+3a, this forces s5 to
be 1 also, because only numbers are divided 3 or 1 and 3, 3 is
not of this form.
(Refer Slide Time: 06:07)

So, in other words, G, remember the whole point is that, we

know nothing about G, other than that it is order of 15, but using
only that, we say G has exactly one Sylow 3-subgroup and G
has exactly, so, let us call this H, and G has exactly one Sylow
5-subgroup K. So it has exactly one Sylow 3-subgroup and
exactly one Sylow 5-subgroup.

467
And now by second Sylow theorem, H and K are normal in,
remember again I want to stress this again and again, we know
nothing about G, that is whole point. We can conclude this for
any group of order 15, of course G is order 15, nothing more.

So because there is only 1 Sylow 3-subgroup, any two Sylow

subgroups, 3-sub-groups are conjugate, H must be normal in G;
similarly K must be normal in G. So note that H is a Sylow 3-
subgroup, this implies H has order 3, but that means H is
isomorphic to Z/3Z, okay. This is a consequence of first
isomorphism theorem way back. If you have a cyclic group of
certain order, first of all H is a prime order group, so that must
be cyclic and any cyclic group of order 3 must be isomorphic to
Z/3Z.
(Refer Slide Time: 08:01)

Similarly, K is the isomorphic to Z/3Z, Z/5Z. Now, we claim,

consider the map from H x K to G. So what is this map, I send,
so this is the cartesian product, (h,k), it goes to h times k, so h
and k are elements of capital H, and Capital K right, but capital
K, and capital H are subgroups of G so, small h and small k are
elements of capital G, so I can multiply them, so I claim that this
is an isomorphism.

And remember that, this proves that any group, any arbitrary
group of order 15 is isomorphic to Z/3Z x Z/5Z. So, to prove
that phi is an isomorphism, we need to show several things.
(Refer Slide Time: 09:02)

468
First, phi is a homomorphism; so before that, we first show, I
want to show that, H intersection K is e, why is this? This is
easy, because H intersection K is a subgroup of K, of H or K, it
is a subgroup of H, by Lagrange‘s theorem, order of H
intersection K divides order of H, which is 3 okay, so in other
words, order of H is intersection K is 1, or order of H
intersection K is 3.
(Refer slide Time: 10:06)

If it is 1, H intersection K must be identity, because identity is

certainly in H and K, so it is in the intersection. But if it is 3,
then H intersection K is okay, but then, we also use that, now,
assume this okay. We also know H intersection k is subgroup of
K okay, so order of H intersection K divides order of K is 5, so
this cannot happen right, so this cannot be 3, because 3 divides
5, so order of H intersection is a number divides 5 and 3, so it
has be 1, okay. This is going to be useful to us. H intersection K
is just the identity element.

Now, let us show one by one, that it is a homomorphism, then it

is 1-1 and then it is onto. Why is it a homomorphism, what we
need to check here, to check, what we need to check to, check
something is a homomorphism, we need to check that, phi of
(H1,K1) times (H2 ,K2),

(Refer Slide Time: 11:23)

469
we want to check this is equal to phi of (H1,K1) times phi of
(H2,K2) , right. So we want to check that, in order for something
to be a homomorphism you multiply in the left hand side group,
apply the map, or your apply the map and then multiply, you get
the same answer, multiply first, apply phi or apply a phi first and
multiply.

So what is this? Remember the product is component-wise, and

this is H1K1 phi of (H1,K1) is that, phi of (H2,K2) is H2K2 , this
is H1H2K1K2 and this is H1K1H2K2. So now the question is are
this equal? So now, you must remember that I am not necessary
assuming that, G is abelian, if G is abelian, of course it will
follow immediately that these are equal, but I am not assuming
that, however I claim that,
(Refer Slide Time: 12:37)

we have, if h is in H, h is in K, then, they commute. I claim that

they commute, why is this? Let us right two elements as follows,
so we have (hkh inverse) times k inverse, is equal to h(kh
inverse k inverse). So, I am not doing anything here, I am just
grouping them differently in a group, that gives you the same
answer, so h k hinverse, k inverse, hk h inverse k inverse. But
now, remember that, this is in K, why is that because K is
normal in G, small k is an element of Capital K right, and small
h is in H, that is irrelevant, it is in capital G.

470
So hk h inverse will be in K, so, here is where you know, both
second and third Sylow theorems are being used. To show that
K is the only Sylow 5 sub-group, we needed Sylow third
theorem and to know that, it is normal, we needed second Sylow
theorem.

So this is in K of course, this is also in K, so this wholee thing in

K. But now, let us look at this, this because H is normal that is in
H, right, because H is in normal, h inverse is in H, h is in G, this
is in H, this is in H, the product is in H, but we have already
argue that,
(Refer Slide Time: 14:19)

K intersection H is in empty, sorry just the identity element.

This means this element, is the identity element, but that means,
hk is kh, because you can multiply by first k and then by h. So,
while it is true that G is not necessary abelian, elements of H,
and elements of K commute with each other.

Now let us come back here, we have H1 H2 K1K2. This can be

written as H1 interchange these two, which gives this, okay.

So, phi is a homomorphism, if you can inter change H 2, K1, you

are done. So this shows that, phi is a homomorphism, we are

471
claiming that is an isomorphism, so let us prove that phi is 1-1.

Suppose that, phi of H1 K1,sorry, phi of (H,K), so can do this,

phi of (H2 ,K2). But what is this, this implies, so if two things
map to the same thing, I want to conclude those two things are
equal, so this is H1K1= H2K2 . That is the map phi, it takes
(H1,K1) to H1 times K1, but that means, H2 inverse H1 is K2K1
inverse, but this is in H, this in K, again use that H and K have
nothing in common.

This implies that, if this element seems to be in both H and K, so

this is e, this means, similarly K2 K1 inverse e. This implies
from H1 is H2 , this implies K1 =K2 okay so phi is 1-1.
(Refer Slide Time: 16:25)

Now why is phi is onto? Remember again phi is a map from H x

K to G, it sends (h,k) to h times k, okay. So what we have shown
is, I claim that H x K has 15 elements.

So suppose for the moment you grant me this claim, if it has 15

elements, it is a 1-1 map to G, which is also 15 elements, it
must be bijective, it must be onto. In fact, we claim two, why
does it follow that it has 15 elements? This follows, this implies
first of all phi is onto right, because phi is 1-1, and G has also 15
elements. So the claim completes the argument that phi is an
isomorphism. You have a 15 element set mapping to a 15
element set, but it is 1-1, so it must be onto also so, this proves

472
the proposition.

The claim that phi is an isomorphism, so why does H cross K

have 15 elements?

(Refer Slide Time: 18:00)

So first we show that, we can show that H x K is a subgroup

okay. I will leave this an exercise for you, the proof is exactly as
before, using the same idea that, if you have HK=KH, you can
use that, okay, so it is an easy exercise, so this is easy, for you.
Then I claim that order of H x K, order of H x K, not claim, this
follows, this is Lagrange’s theorem, this divides 15, Lagrange’s
theorem says that it divides 15.

But certainly order of H x K is greater than 5, because K has

already 5 elements and you are multiplying by H, and these are
different elements, it must be at least 5, so the only number that
divides 15 and bigger than 5 is 15, so that shows that, H x K is
15, has 15 elements and hence, H x K.

(Refer Slide Time: 19:19)

Thus phi from H x K to G is an isomorphism right, so it is an

isomorphism. Now, we have, so we showed that, we recall that
H is isomorphic to Z/3Z, and K is isomorphic to Z/5Z, so G is

473
isomorphic to Z/3Z, isomorphisms are preserved by
composition. So G is isomorphic to Z/3Z x Z/5Z for any group
of order 15.
(Refer Slide Time: 20:04)

So in particular Z/15Z, which is a group of order 15, is

isomorphic to Z/3Z x Z/5Z, and which is further isomorphic to
any group of order of G.

So any group, being an isomorphism is a transitive relation so, G

is isomorphic to Z/15Z, and hence G is cyclic. This proves the
proposition. We showed that any group of order 15 is cyclic.
Hence, this is a very good application of Sylow theorems.

We have no way of proving that an arbitrary 15 order group is

cyclic, I mean how do we do that, there could be lots of groups,
we do know the structure of those groups but the strength and
power of the Sylow theorem is that, we can conclude that any
group of order 15 is cyclic.
(Refer Slide Time: 21:12)

Now, let me as further examples of Sylow theorems, let me look

at some problems. Show that any group, a group of order 100
has a normal subgroup of order 25. So, this is an easy

474
application of the Sylow theorems again, so we recall that, we
know that order of G is 100, which is 52 times, 4, so if you want
to look at, first of all there is, a Sylow 5-sub –group of G has
order what? has order 52, because 52 is the largest power of 5
appearing in 100 so, we know by the first Sylow theorem, we
know that there is a subgroup of order 5, but why is there normal
subgroup of order 5. By first Sylow theorem, we know that G
has a subgroup of order 25, but why normal? Why should there
be a normal subgroup of order 25? For use, to prove that we
have to use the remaining two Sylow theorems.
(Refer Slide Time: 23:03)

Let’s say S25 is the number of Sylow 5-subgroups, so I am using

S25 because to emphasise the fact that Sylow 5-subgroups have
order 25.

By Sylow third theorem, by Sylow III, S 25 divides 4, and

S25=1+5a. Remember that, the prime plays a role here, 5 is the
prime that is relevant over here, so S25 divides 4, and S25= 1 +5a,
but S25 dividing 4 can be 1 or 2 or 4. 1 of course is possibility,
but if it is 2, it is not of this form right, it is not when you divide
by 5 the remainder is not 1, similarly it cannot be 5, 4. So S25
has to be 1, so there is only 1 Sylow 5-subgroup and by Sylow
II, it must be a normal right so, that solves the problem.
(Refer Slide Time: 24:37)

475
Just as an aside, let us look at Sylow 2-subgroups; 2 is the other
prime dividing 100, have order 4 right, 100 is 4 times 25, so
Sylow 2-subgroups have order 4.

So let us say S4 is the number of Sylow 2-subgroups so, S4

divides 25 and S4 is 1+2a okay.

Now, what are the possibilities? If it divides 4, it is 1, sorry, it

divides 25, either 1 or 5 or 25. Now 1 is certainly a possibility
but even 5 is a possibility or 25 is a possibility. So without
knowing further about the group, we cannot restrict choices any
more. So the number of Sylow 2-subgroups is either 1 or 5 or
25. So we do know which one actually happens.

So this tells you both the strength and the limitation of Sylow
theorems, in the case of Sylow 5-subgroups here, we are able to
conclude one is the only option and it must be normal, but for
Sylow 2-subgroups, there are three possibilities. And Sylow
theorems themselves do not tell us how to eliminate any of those
possibilities.
(Refer Slide Time: 26:02)

Let me do one more problem along the same lines, okay. Let us
say p and q are distinct primes and suppose p strictly less than
Q. Let G be a group of order pq, show that, G has a normal

476
subgroup of order q, okay.

So I will solve it very quickly, the proof is very simple, solution

is very simple and similar to previous problem. What is the
idea? We want show that there is a exactly one Sylow q-
subgroup, so, Sylow q-subgroups of G have order q, it is not q 2
or q3 right, because q is the largest power of q appearing in order
of G. So they have, Sylow q-subgroups have order q and always
we denote Sq to be the number of Sylow q-subgroups. We know
that Sq divides, Sq divides p and Sq is 1+aq, right.

But let us see. If Sq divides p, Sq must be 1 or p, but can it be 1?

Of course it can be 1, but it can be p? Can p be written as 1+aq?
It cannot be, because p is smaller than q, p is smaller than q
means p cannot be return as 1+ aq, because 1+aq is more than q.

So G has only one Sylow q-subgroup right. We are using the

third Sylow theorem here. G has only one Sylow q-subgroup, let
us say H. And Sylow II implies H is normal in G, because any
two Sylow q-subgroups are conjugate, so if you take H and
conjugate it, it is another Sylow q-subgroup because it is a
subgroup and it has q elements. But because there is only one
Sylow q-subgroup, that must be H, so all conjugates of H are H,
so H is normal in G. So this is the solution.

Okay, these two problems gave you some idea of how to apply
these Sylow theorems and how why they are very powerful,
okay. So, I will stop this video here, in one more video I will do
some more examples and problems which illustrate how to use

477
Sylow theorems. Thank you.

Online Editing and Post Production

Karthik
Ravichandran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh Kumar
Web Studio Team
Anitha
Bharathi
Catherine
Clifford
Deepthi
Divya
Gayathri
Gokulsekar
Halid
Hemavathy
Jagadeeshwaran
Jayanthi
Kamala
Lakshmipriya
Libin
Maria Neeta
Mohana
Mohana Sundari
MuraliKrishnan

478
 Nivetha
Parkavi
PoonKuzhale
Poornika
Premkumar
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirely
Sorna
Subash
Suriyaprakash
Vinothini
Executive Producer
Kannan Krishnamurty
NPTEL Co-ordinators
Prof. Andrew Thangaraj
Prof. Prathap Haridoss
IIT Madras Production
Funded by
Departmental of Higher Education
Ministry of Human Resource Development
Government of India
www.nptel.ac.in

479
480
 NPTEL
NPTEL ONLINE COURSE
Introducing to Abstract Group Theory
Module 09
Lecture 45 – “Problems 10”
PROF. KRISHNA HANUMANTHU
CHENNAI MATHEMATICAL INSTITUTE

Okay, so, in this video I will continue my applications of Sylow

theorems and problems based on Sylow theorems. But before
that let me do actually a couple of small results that don’t use
Sylow theorem. But, they are useful applications for group
actions. So, let’s say this is a problem I want to solve today.
(Refer Slide Time 00:33)

Let’s say G is a group and p is a prime number and let’s say G

has order pe. Okay, in other words order of G is a power of p and
let’s say e is at least 1. Okay, show that the center of G is non-
trivial. Okay so, I will explain what this means.

So recall the center of G, which is denoted by Z(G), is all

elements of G which commute with all other elements. So, the
problem is asking us to show it is non-trivial.

So, we want to show order of Z(G) is at least 2. Okay, so

remember this need not be the case. I will only say this, you can
check this on your own. If G is S3 for example, the symmetric
group on 3 letters, center is thriven. Of course, the symmetric

481
group doesn’t have this property that its order is a prime power.
So, if you have this, prime power groups have non-trivial order.
So, for this we consider the action of G on itself by conjugation.

Okay, so this is something we have seen before. So, in

particular, we look at class equation. What is the class equation?
It says that cardinality of, order of G, which I am assuming is Pe,
is the sum of orders of conjugacy classes. But remember one
conjugacy class is just {e}, this is the orbit of e. Conjugacy
classes are orbits of elements under conjugation action. So, the
class equation looks like, pe=1+|C2|+|C3|+…+|Ck|.

(Refer Slide Time 02:47)

Now we further apply group actions. So, what does counting

formula say? Counting formula says order of G is stabilizer of
some element times orbit of that element. Again this is p e so we
can certainly say that the size of a conjugacy class or size of an
element, orbit of an element divides pe. But what are the
numbers that divide pe? So, orbit of g must have size pi for i = 0,
1, 2 up to e, it can be zero, of course.

Now let’s go back to the conjugacy class equation. You have

1+|C2|+|C3|+…+|Ck|; using the counting formula I can say that,
each of these is a power of p even 1 is a power of p. So, this is
some pa2 +pa3+…+pak , correct.
(Refer Slide Time 04:33)

482
Suppose a2 is positive, a3 is positive, ak is positive, suppose
they are all positive. Then, pe will be 1+pa2 +pa3+…+pak , and
this whole thing is divisible by p, because if they are all positive.

That means p divides 1 because p divides pe, p divides this sum,

so p divides their difference which is 1. But this is absurd.

But, where did we go wrong? We went wrong in assuming that

all the ais are positive. So, some ai is equal to 0. So, say a2 is 0.
That means what? |C2| =1.
(Refer Slide Time 05:39)

But what is C2? So a2 remember is the, |C2| is pa2, a2=0. So, |C2|
is 1. This implies C2 which is the conjugacy class or orbit of
some element g has only one element. But, what is orbit of g? It
is all elements like this. But of course orbit definitely contains g.
But that means aga-1 is g for all a in G. That means, ag = ga for
all a in G; that means a belongs to the center. Sorry, this is not
‘a’ that I can conclude is in the center, a is varying so I can
conclude g is in the center and remember g is different from e.

(Refer Slide Time 06:42)

Because I am taking the second thing that appears in the class

equation, these are distinct orbits, e already is taken care of
when we wrote 1. So C2 being the order of g and g must be

483
different from e. So we have produced an element g in the center
which is different from e. So that concludes that the center is
non-trivial. Okay, so that solves the problem.
(Refer Slide Time 07:30)

So, if you have a group whose order is prime power, such

groups are called p-groups, we have shown that the center is
non-trivial. So, p groups have non-trivial center.

Now I am going to consider a special kind of p-group. Suppose

G has order p or p2, where p is prime then G is abelian. So it is
not true that p-groups are abelian, all we know is that their
center is non-trivial. So, there are non-identity elements in the
center. But if it is either p or p2 the center is everything, in other
words is an abelian.

So, if |G|=p then G is actually a cyclic group, implies G is

abelian. There is nothing new here we have seen this before.
(Refer Slide Time08:35)

So assume |G|=p2. So we want to show G is abelian.

Equivalently we want to show Z (G) = G. Note that by the
previous problem we know that the center is non-trivial. So,
center must be bigger than just the identity. But now we want to
show that it is all of G.

What are the possibilities for the center? So, center is a subgroup

484
(e) ≤ Z (G) ≤ G, which is between (e) and G. By the previous
problem it is not equal to (e), so it is either G or it is strictly
between. So suppose, if possible, Z G) ≠ G.

So, you should think for a minute if it is not clear to you, why
the center being all of G implies G is abelian. This is just the
definition of what a center is.

So suppose the center is not equal to G. So choose x ϵ G, x not

in the center Z(G), right. If x is not in the center, if center is not
all of G there must be something that is not in the center. Let’s
consider the conjugacy class of x, that means C(x) = {gxg-1 | g ϵ
G}. Okay so, this is called the centralizer of x, recall, and it is a
subgroup. And I will let you do that easily, check that it is a
subgroup of G. Because identity certainly has this property, if
two things have that property their product has this property and
if something has this property inverse also has this property.
(Refer Slide Time11:00)

Now, this is easy check. It is also easy to check that center will
always be contained in the centralizer, because center, remember
is all elements which commute with everything. So, if
something commutes with, sorry this is actually, I made a
mistake here so, actually let me erase all this, I got confused. I
am not looking at the conjugacy class. I’m looking at the
centralizer.

485
Consider the centralizer of x, I will call it C(x). This is by
definition {g ϵ G| gxg-1 =x}. Okay so, this is a subgroup of G
and this contains the center Z(G). That is what I want to say.
Why is this? This is all group elements which commute with
everything. C(x) is the group elements which commute with x
so, Z (G) contains C(x). So, what do we have?

We have (e), we have Z (G), we have C(x), we have G. The

reason we need p2 is this, so this is order 1, this we are assuming
is order p because it is strictly between G and ZG and this is p 2
and C(x).
(Refer Slide Time 12:49)

Remember, C(x) contains x because x certainly has this

property: xxx-1 is x. So, Z(G) and Z(G) does not contain x, that
is our assumption. So this implies Z(G) is contained but not
equal to centralizer. So, this must be p2.

That means, so C(x) = G, all of G itself, that means g x = x g for

all g ϵ G right, that is the centralizer. It is all elements which
have g x = x g, but that is all of G, so g x = x g for all G.

But this in turn means that x ϵ Z (G), this is a contradiction right,

because we chose x to be not in Z (G). So this shows that any
group of order p2 or any group of order p is abelian.
(Refer Slide Time14:13)

486
So, now using these two facts which do not require Sylow
theorems at all, these two problems, I am going to do a third
problem which does use Sylow theorem.

So, let’s say G is a finite group and p divides |G| and as always p
is a prime number. Okay, in fact, suppose that p e divides |G| then
G has a subgroup of order pe. Okay so this is the problem.

This looks very much like first Sylow theorem but with some
difference, the first Sylow theorem is the key statement we will
use but we have to do some more work because the first Sylow
theorem says that, you take the largest power of p that divides
|G| then there is a subgroup of that order. Here, I am not
assuming that p^e is the largest power. So, how do we do this?
So, first without loss of the generality we can assume G is a p-
group, that is |G| =pn.

(Refer Slide Time15:56)

Why is this? Meaning why can we assume this? So, write G as

some, I think maybe I should not use e here. Let’s say, suppose
G is a finite group and pi divides G then G has a subgroup power
of pi.
(Refer Slide Time 16: 20)

487
And here I can assume that I claim this. So, write like this, like
we do in the Sylow theorem, so p does not divide m. Sylow I
implies G has a subgroup order pe, say H is the subgroup of G of
pe . Now, if i ≤ e, if we show, take i ≤ e, if we show H has a
subgroup of order i, then G also has a subgroup right, the same
subgroup which is the subgroup of H is also a subgroup of order
pi for G also. So, it’s enough to show that H contains a subgroup
of order pi. So, we can restrict our attention and this, in order to
make this assumption we need the Sylow theorem, in the rest of
the proof we do not need. So Sylow theorem is needed for this
reason.
(Refer Slide Time 17: 52)

So, assume now that G is, we can from now on assume |G|=p e
and i ≤ e, we want to show G has a subgroup of order p i.

Okay, just to quickly recap why we can make this assumption: G

need not be a p-group but it will have, its order will decompose
like this, Sylow I says that G has a subgroup of order p e and if
you prove the problem for that subgroup, we have solved the
problem for the given subgroup, given group G. So we might as
well assume that G is a group of order p e.

Now by a corollary of first Sylow theorem, we can assume that.

No, not corollary of the first Sylow theorem, by the first
problem of this video, the center of the group is non-trivial,
right. We have shown that a p-group has non-trivial center. So,
take, it has non identity element. So I can always choose an
element, let’s call this x, of the center such that order of (x) = p.

488
And in order to do this I used the corollary to the first Sylow
theorem. Remember there is an element of the center which is
not identity, order of the center is also a power of p, in other
words, the center is also a p-group because order of the center
divides pe. So, the order of the center is also a power of p. So an
element will have order a power of p. So, by taking a suitable
power of it, we can assume that it has order p. So look at
Corollary to Sylow I that we have done in that video for
producing such an x and now let H be the subgroup.

So I am starting with a new notation here. Whatever I called

earlier H is gone now so, G is a group of order pe and H be the
subgroup generated by x. So, H is the subgroup generated by x
so |H| = p, of course, because order of x is p.
(Refer Slide Time 20:40)

More than that, H is in fact normal in G. Why is that? Because x

is a central element, x is in the center. So how do you show
something is normal? Take an arbitrary group element g and an
arbitrary ring element, it will be some x n and multiply by g-1.
But remember x is in the center right, we have chosen an
element x of the center. So, xn also is in the center so xn gg-1 =
xn, this is of course in H. So H is normal in G.
(Refer Slide Time 21:28)

489
So, in particular we can consider the quotient group G/H. Right,
if you have a normal subgroup of a group you can consider the
quotient group and look at the map from G to G/H, the natural
map so, g going to the coset gH.

What is the cardinality of, order of |G/H|? By the counting

formula this is the ratio, pe this is p so, so this is pe-1, okay. Now,
by induction, so I am going to basically solve this problem by
induction. So we use induction. |G/H| = p e-1 which is less than
pe.

So, we can assume by induction that G/H has a subgroup, say

K1, such that the |K1| = pi-1. So, I am going to construct, my goal
is to construct a subgroup of order pi for G.
(Refer Slide Time 23:05)

I am going to construct, I am going to use induction hypothesis

to suppose that G/H, which is a group of order less than p e has a
group of order, subgroup of order pi-1.

So, what is the picture: we have G, G/H, K1, which is the

subgroup, this has order pi-1, this has order pe-1, this has order pe.
How do I get, so this is the picture how do I get the K that I we
want, the group of order subgroup of order p i? Okay so,
naturally your guess would be, I take φ-1 (K1), call that K.

490
So, let K be phi inverse (K1). It is an easy exercise that I may
have discussed when I talked about homomorphisms, inverse
image of a subgroup is a subgroup. So, K is a subgroup of G.
We claim that K is the desired subgroup, we claim that |K|=p i.

So remember our goal is to construct, given than G has order p e

and i ≤ e, we want to show G has a subgroup of order p i. We use
induction and go to a smaller group which is G/H and apply
induction to that to guarantee that there is a K1 for, which is a
subgroup of G/H which has order pi-1.

(Refer Slide Time 24:45)

And I am simply taking the inverse image, K will be the inverse

image. So, it is a subgroup, that is trivial. We want to show that
in fact, its order is pi. So, why is that? This is very easy now. So
remember, we have G to G/H, K is here and K to K1, we have a
map.
(Refer Slide Time 25:08)

Because K, this is phi, this is phi restricted K. And because K is

the inverse image of K1, K maps to K1. So, what is the kernel of
this map? Remember what is the kernel of phi first? Kernel of
phi is of course H, right. kernel is the inverse image of identity
element. So K1 contains the identity element, so its inverse
image will contain all of the kernel, so K contains H. We have

491
this also, so kernel of phi restricted to K, I claim, is also H.
Because how can something go to 0 here or the identity element
here?

If it goes to identity element here, it already goes to identity

element here, so it is in H. So, if something is in H it is already
in K, so kernel of this is H. And of course phi restricted to K is
also onto, it is onto right, because phi itself is onto. So, this is
also onto.

So forget this now, focus your attention on this map and apply
counting formula. It says that, or the first isomorphism theorem
really, first isomorphism theorem says that K mod kernel of this
map, which I am denoting by phi restricted to K, is isomorphic
to K1.

But that is same as, because kernel is H, this implies that now
the counting formula says that |K| by |H| is |K1|, which remember
is pi-1. So |K| = p i-1 times |H|, but the order of H is p, okay.

(Refer Slide Time27:04)

So thus we have produced a subgroup K of G of order p i. Okay,

remember this was the problem. Problem was to show that if
you have a group G such that pi divides the order of G, then
there is a subgroup of order pi. So now, first using the first
Sylow theorem, we have reduced to the case that G has order p
power e; that was the crucial restriction and that needed Sylow’s

492
first theorem.

Then we used the fact that, for such p-groups the center is non-
trivial and to get hold of this x which made all the proof work.
And then, we go to G mod the subgroup generated by x, use
induction to show that that has the right subgroup and we take
the inverse image and then a quick application of first
isomorphism theorem and counting formula says that the inverse
image has the desired property. So, this proves the problem and
this completes the video and the course.

I wanted to cover the standard things that are done in a group

theory course: so we have done Lagrange’s theorem, cosets,
quotient groups, Cauchy’s theorem, group actions which is the
most important topic here and using that we have done the
Sylow theorems. So this is the material that is covered in any
standard course in group theory and if you understand this
course fully, then you know the group theory that is expected in
a B.Sc or a M.Sc course. Thank you.

Online Editing and Post Production

Karthik
Ravichandran
Mohanarangan
Sribalaji
Komathi
Vignesh
Mahesh Kumar
Web Studio Team
Anitha

493
 Bharathi
Catherine
Clifford
Deepthi
Dhivya
Divya
Gayathri
Gokulsekar
Halid
Hemavathy
Jagadeeshwaran
Jayanthi
Kamala
Laksmipriya
Libin
Madhu
Maria Neeta
Mohana
Mohana Sundari
Muralikrishnan
Nivetha
Parkavi
Poonkuzhale
Poornika
Premkumar
Ragavi
Raja
Renuka
Saravanan
Sathya
Shirley

494
 Sorna
Subash
Suriyaprakash
Vinothini
Executive Producer
Kannan Krishnamurthy
NPTEL Co-ordinators
Prof. Andrew Thangaraj
Prof. Prathap Haridoss
IIT Madras Production
Funded by
Department of Higher Education
Ministry of Human Resource Development
Government of India
www.nptel.as.in

495