NOTES ON GROUP THEORY

Abstract. These are the notes prepared for the course MTH 751 to
be offered to the PhD students at IIT Kanpur.

Contents
1. Binary Structure
2. Group Structure
3. Group Actions
4. Fundamental Theorem of Group Actions
5. Applications
5.1. A Theorem of Lagrange
5.2. A Counting Principle
5.3. Cayleys Theorem
5.4. The Class Equation
5.5. Cauchys Theorem
5.6. First Sylow Theorem
5.7. Second Sylow Theorem
5.8. Third Sylow Theorem
6. Structure Theorem for Finite Abelian Groups
References

2
5
13
15
17
17
17
18
18
19
21
21
22
24
26

NOTES ON GROUP THEORY

1. Binary Structure
Let S be a set. We denote by S S the set of ordered pairs (a, b), where
a, b S. Thus the ordered pairs (a, b) and (b, a) represent distinct elements
of S S unless a = b.
A binary operation ? on S is a function from S S into S. Thus for every
(a, b) S S, the binary operation ? assigns a unique element a ? b of S. If
this happens, then we say that the pair (S, ?) is a binary structure.
Let us understand the above notion through examples.
Example 1.1 : We follow the standard notations to denote the set of natural numbers, integers, rationals, reals, complex numbers by N, Z, Q, R, C
respectively. If S is one of the sets above, then S stands for S \ {0}.
(1)
(2)
(3)
(4)

Addition (resp. multiplication) is a binary operation on Z (resp. Q).

Division is not a binary operation on Z .
Subtraction is a binary operation on Z but not on N.
Division is a binary operation on R (resp. C ).

As seen in (3), it may happen that a ? b

/ A for some a, b A.
Let (S, ?) be a binary structure. Let A be a subset of a set S. We say
that ? is an induced binary operation on A if a ? b A for every a, b A.
Exercise 1.2 : Let O denote the set of odd integers. Verify that the multiplication on Z is an induced binary operation on O, however, addition is
not so.
Let us see some geometric examples of binary structures.
Example 1.3 : Let T denote the unit circle. Consider the binary operation
of multiplication from T T into T. Note that the action
(1.1)

(z, w) z w

can be interpreted as rotation of z about the origin through the angle arg(w)
in the anticlockwise direction.
As an another interesting example of a binary operation, consider the
binary operation of multiplication on an annulus centered at the origin.
One may use the polar co-ordinates to interpret the action (0.1) as rotation
of z about the origin through the angle arg(w) in the anticlockwise direction
followed by a dilation of magnitude |w|.
Exercise 1.4 : Let A(r, R) denote the annulus centered at the origin with
inner radius r and outer radius R, where 0 r < R . Find all values of
r and R for which (A(r, R), ) is a binary structure.

Hint. If r
< 1 then r = 0 (Use: r < r if 0 < r < 1). If R > 1 then
R = (Use: R < R if 1 < R < ).

NOTES ON GROUP THEORY

Exercise 1.5 : Let L denote a line passing through the origin in the complex
plane. Verify that the multiplication on the plane is not an induced binary
operation on L.
A binary structure (S, ?) is associative if x ? (y ? z) = (x ? y) ? z for every
x, y, z S. We say that (S, ?) is abelian if x ? y = y ? x for every x, y S.
Exercise 1.6 : Let Mn (R) denote the set of nn matrices with real entries.
Verify that the matrix multiplication is a binary operation on Mn (R).
Verify further the following:
(1) is associative.
(2) is abelian iff n = 1.
Let (S, ?) be a binary structure. We say that e S is identity for S if
e ? s = s = s ? e for every s S.
In general, (S, ?) may not have an identity. For example, the infinite
interval (1, ) with multiplication is a binary structure without identity.
Proposition 1.7. Identity of a binary structure, if exists, is unique.
Proof. The proof is a subtle usage of the definition of the binary operation.
Suppose (S, ?) has two identities e and e0 . By the very definition of the binary
operation, the pair (e, e0 ) assigned to a unique element e ? e0 . However, e ? e0
equals e if e0 is treated as identity, and e0 if e is treated as identity. Thus we
obtain e0 = e as desired.

Before we discuss the isomorphism between two binary structures, it is
necessary to recall the notion of isomorphism between sets. We say that
two sets S and T are isomorphic if there exists a bijection from S onto T.
Recall that Z and Q are isomorphic.
We say that two binary structures (S, ?) and (T, ) are isomorphic if there
exists a bijection : S T, which preserves the binary operations:
(a ? b) = (a) (b) for all a, b S.
We will refer to as the isomorphism between (S, ?) and (T, ).
Remark 1.8 : The set-theoretic inverse 1 of is an isomorphism between
(T, ) and (S, ?).
It is not always easy to decide whether or not given binary structures are
isomorphic. The following two tests are quite handy for this purpose.
Exercise 1.9 : Suppose the binary structures (S, ?) and (T, ) are isomorphic. Show that if (S, ?) is abelian (resp. associative) then so is (T, ).
Note that the binary structures (R, ) and (M2 (R), ) are not isomorphic.
Proposition 1.10. Suppose there exists an isomorphism between the binary structures (S, ?) and (T, ). Fix a S. Then the following is true:

NOTES ON GROUP THEORY

(1) The equation x?x = b has a solution in S iff the equation xx = (b)
has a solution in T.
(2) There exists a bijection from the solution set S of x ? x = b onto the
solution set T of x x = (b).
Proof. If x0 S is a solution of the equation x ? x = b then (x0 ) T is
a solution of the equation x x = (b). The converse follows from Remark
1.8. Since : S T given by (x0 ) = (x0 ) is a bijection, the remaining
part follows.

Example 1.11 : Consider the binary structures (Z, +) and (Q, +). We already recorded that Z and Q are isomorphic. The natural question arises
whether (Z, +) and (Q, +) are isomorphic ?
Let us examine the equation x + x = 1. Note that the solution set of
x + x = 1 in Z is empty. On the other hand, the solution set of x + x = 1
in Q equals {1/2}. By Proposition 1.10, the binary structures (Z, +) and
(Q, +) can never be isomorphic.
Exercise 1.12 : Whether the following binary structures are isomorphic.
Justify your answer.
(1) (Z, +) and (N, +).
(2) (C, ) and (R, ).
(3) (C, ) and (C , ).
Exercise 1.13 : Consider C and T as topological spaces with the topology inherited from the complex plane. Show that there does not exist a
continuous isomorphism from (C , ) onto (T, ).
In view of the last exercise, one may ask: Is it true that (C , ) and (T, )
are isomorphic? The answer is No (refer to [2]).
Exercise 1.14 : Show that there exists no isomorphism between the
binary structures (M2 (R), ) and (M3 (R), ) such that (I) = I.
Hint. Consider the equation A2 = I for invertible solutions A.
Recall that a matrix A Mn (R) is orthogonal if At A = I. Note that
A Mn (R) is orthogonal if and only if A preserves the euclidean distance,
that is, kAX AY k2 = kX Y k2 for every X, Y Rn , where k k2 denotes
the euclidean norm on Rn .
Exercise 1.15 : Prove that any orthogonal matrix in M2 (R) is either a
rotation R about the origin with angle of rotation or a reflection
about the line passing through origin making an angle /2, where




cos sin
cos sin
(1.2)
R =
, =
.
sin cos
sin cos
Hint. Any unit vector in R2 is of the form (sin , cos ) for some R.

NOTES ON GROUP THEORY

Here is an example of geometric nature.

Example 1.16 : Let denote an equilateral triangle in the plane with
origin as the centroid. For example,
one may take triangle with vertices

(1, 0), (1/2, 3/2), (1/2, 3/2). By symmetry of , we understand orthogonal 2 2 matrix A in M2 (R) such that A() = . Consider the set
S3 all symmetries of . It is easy to see that (S3 , ) is a binary structure.
To understand this binary structure, we need a bit of plane geometry.
By Exercise 1.15, any element of S3 is a composition of rotations about
the origin and reflections about a line passing through origin. Since is
an equilateral triangle with centroid 0, a rotation belongs to S if and only
if the angle of rotation is either 2/3, 4/3, 2. Similarly, since the axes
of symmetry of are precisely the lines passing through the origin and
mid-point of sides of , a reflection belongs to S3 if and only if the line of
reflection is one of the axes of symmetry of . It follows that S3 consists
exactly six elements; 3 rotations and 3 reflections.
Exercise 1.17 : Describe all symmetries of a regular polygon in the plane
with origin as the centroid.
2. Group Structure
In the last section, we discussed many examples of binary structures (S, ?).
We also observed that there are some distinguished binary structures,
namely, unital pairs (S, ?) for which the binary operation is associative. It
is desirable to pay more attention to such structures. An axiomatic approach
is often convenient for such studies.
Definition 2.1 : A binary structure (G, ?) is a group if
(1) (Associativity) For all a, b, c G, we have (a ? b) ? c = a ? (b ? c).
(2) (Existence of Identity) There exists e G such that e ? a = a = a ? e
for all a G.
(3) (Existence of Inverse) For all a G, there exists a1 G (depending,
of course, on a) such that a ? a1 = e = a1 ? a.
We say that a group structure (G, ?) is abelian if
(4) (Commutativity) For all a, b G, we have a ? b = b ? a.
Remark 2.2 : Note that the inverse of the identity is the identity itself.
If there is no confusion, we will suppress the binary operation ?.
Note that (R, +) and (R , ) are group structures.
Example 2.3 : For a positive number c, consider the open interval G =
(c, c) of real numbers. For x, y G, define
x ? y :=

x+y
.
1 + xy/c2

NOTES ON GROUP THEORY

Notice that 1 + xy/c2 > 0 for any x, y G, so that x ? y R. To see that ?

is a binary operation, we should check that c < x ? y < c if c < x, y < c.
Note that |x y| < c iff c|x + y| < c2 + xy. If x + y 0 then by |x y| < c
iff (c x)(c y) > 0. Similarly, one can treat the case x + y < 0.
Clearly, 0 is the identity for G and the inverse of x is x. It is easy to
see that ? is associative and commutative. Thus (G, ?) is an abelian group
structure. This example arises in Special Relativity.
Note that (C , ) is a group structure.
Example 2.4 : For a positive integer n, let In denote the set of nth roots
of unity:
(2.3)

In := { C : n = 1}.

By the fundamental theorem of algebra, In consists of exactly n elements

including 1. Geometrically, In consists of the vertices of the regular polygon
with n edges and with centroid the origin.
The binary structure (In , ) is indeed a group structure. To see this, note
first that In C . Now In admits the inverse 1/. Associativity and
commutativity of In follows from that of C .
Exercise 2.5 : Let I := n1 In . Show that (I, ) is a group structure.
Exercise 2.6 : Fill in the blanks and justify:
(1) Let A(r, R) denote the annulus of inner-radius r and outer-radius R
with the assumption that 0 r < R . Then the binary structure
(A(r, R), ) is a group structure if and only if r = and R = .
(2) Let X be a subset of the group structure (C , ). Let SX denote the
set of n n matrices A Mn (C) such that the determinant det(A)
of A belongs to X. Then (SX , ) is a structure if and only if (X, )
is a structure.
(3) Let X be a set containing at least two elements and let P (X) denote
the power set of X. Define A ? B (resp. A B) be the symmetric difference (resp. difference) of A and B. Then (P (X), ?) (resp.
(P (X), )) is a structure but not a structure.
The following summarizes some elementary properties of the group.
Proposition 2.7. Let G be a group. Every element of G has a unique
inverse. More generally, for a, b, c G the following statements hold: If
ab = ac then b = c, and if ba = ca then b = c. In particular, the inverse of
a ? b is given by b1 a1 .
Proof. We will only prove that if ab = ac then b = c. Here inverse a1 of
a works as a catalyst. By the definition of binary operation, a1 (ab) and
a1 (ac) define the same element of G. The desired conclusion now follows
from the associativity of G.


NOTES ON GROUP THEORY

Let us see a couple of applications of the last innocent result.

Exercise 2.8 : Let G be a group and let a G. By a2 , we understand
aa. Inductively, we define an for all positive integers n 2. Show that
(ab)n = an bn for all a, b G if and only if G is abelian.
Exercise 2.9 : Let G be a group. Show that if G is non-abelian then G
contains at least 5 elements.
Hint. Observe that there exists x, y G such that e, x, y, xy, yx are
distinct elements of G.
Remark 2.10 : A rather extensive usage of Proposition 2.7 actually shows
that a non-abelian group can not contain 5 elements. We will however
deduce this fact later from a general result.
Example 2.11 : For a positive integer n 3, consider the binary structure
Dn of all symmetries of a regular n-gon with origin as the centroid. As
in Example 1.16, it can be seen that Dn consists of rotations R ( =
0, 2/n, , 2(n 1)/n) and reflections ( = 2/n, 4/n, , 2)(see
(1.2)). Clearly, the rotation by 0, the 2 2 identity matrix, plays the role of
identity for Dn . Either geometrically or algebraically, observe the following:
(1) Rn = R0 = 2 .
(2) R R = .
In particular, the inverse of R is Rn1 and the inverse of is itself.
Thus (Dn , ) forms group structure, which is not abelian.
Remark 2.12 : (D3 , ) is the smallest non-abelian group structure.
Exercise 2.13 : Let A be a set and let PA be the set of bijections f : A A.
Show that (SA , ) is a group structure.
Example 2.14 : For a positive integer n, let A denote the set {1, , n}.
Set Sn := SA . The group structure (Sn , ) is known as the symmetric group.
Note that Sn contains n! elements.
A transformation between two group structures (G, ?) and (G0 , ) is said
to be a group homomorphism if preserves the group operations: (a ? b) =
(a) (b) for all a, b G. We say that (G, ?) and (G0 , ) are isomorphic if
there exists a bijective homomorphism (known as isomorphism) between
(G, ?) and (G0 , ).
Remark 2.15 : Let e and e0 denote the identities of G and G0 respectively.
Let : G G0 be a homomorphism.
(1) Then (e) = e0 . This follows from Proposition 2.7 in view of
(e) e0 = (e) = (e ? e) = (e) (e).

NOTES ON GROUP THEORY

(2) By uniqueness of inverse, for any a G, the inverse of (a) is (a1 ).

Again, whenever there is no confusion, we suppress the symbols ? and .
Example 2.16 : Let m, n be positive integers such that m n. Define
: Sm Sn by () = (m + 1)(m + 2) (n) ( Sm ). Then is an
injective, group homomorphism. Thus is an isomorphism iff m = n.
Let us verify that Sn is abelian if and only if n 2. Clearly, if n = 1, 2
then Sn is abelian. Suppose n 3. Since (S3 ) Sn , it suffices to check
that S3 is non-abelian. To see that, consider = (1, 2, 3) and = (1, 2)(3).
Then = (1, 3)(2) and = (1)(2, 3).
Example 2.17 : Consider the groups (Dn , ) and (Sn , ). Since |Dn | = 6
and |Sn | = n!, (Dn , ) and (Sn , ) can not be isomorphic for n 4. Suppose
n = 3. Define : D3 S3 by setting
(R0 ) = (1)(2)(3), (R2/3 ) = (1, 2, 3), (R4/3 ) = (1, 3, 2),
(/3 ) = (1, 2)(3), (2/3 ) = (1, 3)(2), ( ) = (1)(2, 3).
Clearly, is a bijection. Verify that is a group homomorphism. Likewise
one can see that Dn embeds into Sn .
Exercise 2.18 : Let (Gn , ) denote the smallest group structure consisting
the 2 2 complex matrices
!


2ki
e n
0 1
0
(k = 0, , n 1).
F =
, Ck =
2ki
1 0
0
e n
Show that : Gn Dn governed by
(F ) = /n , (Ck ) = R2k/n (k = 0, , n 1)
defines an isomorphism between Gn and Dn .
Recall that a fractional linear transformation is a rational function fa,b,c,d
of the form az+b
cz+d , where a, b, c, d are complex numbers such that cz + d is
not a multiple of az + b, and |c| + |d| =
6 0.
Example 2.19 : Let F denote the set of all fractional linear transformations. Then (F, ) is a binary structure with identity f1,0,0,1 . Consider the
transformation : F M2 (C) given by


a b
(fa,b,c,d ) =
.
c d
Verify that preserves the binary operations. In particular, the common
notation employed for both matrix multiplication and composition of functions is fully justified. Since is not surjective, (F, ) and (M2 (C), ) are
not isomorphic.

NOTES ON GROUP THEORY

Let (G, ?) be a group structure and let S be a subset of G. We say that

S is a subgroup of G if (S, ?) is a group structure in its own right.
Note that S is a subgroup of G iff
(1) (Binary Structure) ab S for every a, b S.
(2) (Existence of Identity) There exists e0 S such that e0 a = a = ae0
for all a S.
(3) (Existence of Inverse) For all a S, there exists a1 S such that
aa1 = e = a1 a.
Remark 2.20 : Let e denote the identity of G. Since e0 e0 = e0 = ee0 , by
Proposition 2.24, we have e0 = e.
Exercise 2.21 : Let G be a group, a G, and let H be a subgroup of G.
Verify that the following sets are subgroups of G :
(1) (Normalizer of a) N (a) = {g G : ag = ga}.
(2) (Center of G) ZG = {a G : ag = ga for all g G}.
(3) (Conjugate of H) aHa1 = {aha1 : h H}.
Remark 2.22 : Note that a ZG iff N (a) = G.
A permutation Sn is a transposition if either is the identity permutation or there exist integers i < j such that (k) = k for k 6= i, j and
(i) = j, (j) = i, that is, = (j, i). A permutation is said to be even if it
is a composition of even number of transpositions.
Example 2.23 : For a positive integer n 3, consider the set An of even
permutations in Sn . Then (An , ) is a subgroup of Sn . Let us see that |An | =
n!/2. To see that, note that : An Sn \ An given by () = (2, 1) is
a bijection. The group (An , ) is known as the alternating group.
Note that A3 = {(1, 2, 3), (2, 3, 1), (3, 1, 2)}. It follows that A3 is isomorphic to the group of rotational symmetries of an equilateral triangle with
centroid the origin. The elements of (A4 , ) can be realized as rotations of
a tetrahedron.
Recall that a matrix A is diagonalizable if there exists a diagonal matrix
D and an invertible matrix B such that A = BDB 1 . The set of invertible,
diagonalizable 2 2 matrices is not a subgroup of GL2 (R).
Proposition 2.24. Let G and G0 be two group structures. If : G G0 is
homomorphism then ker (resp. ran ) is a subgroup of G (resp. G0 ).
Proof. We will only verify the first statement. By definition, the kernel of
consists of those elements a G for which (a) = e0 , where e0 denotes the
identity of G0 . Now if a, b ker then (ab) = (a)(b) = e0 e0 = e0 . That
is, the binary operation on G is an induced operation on ker . If e denotes
the identity of G then (e) = e0 (Remark 2.15(1)), that is, e ker . Since
(a1 ) = (a)1 (Remark 2.15(2)), if a ker then so is a1 .


10

NOTES ON GROUP THEORY

Exercise 2.25 : Consider the group structures (C , ) and (T, ). Define

: C T by (z) = z/|z| for z C . Show that is a surjective, group
homomorphism. What is the kernel of ?
Let us examine a few instructive examples.
Example 2.26 : Consider the set C[0, 1] of real-valued continuous functions
defined on the interval [0, 1]. Since addition of continuous functions is continuous, (C[0, 1], +) is a group structure. Also, consider the additive group
structure (R, +) of real numbers.
R1
Consider the mapping : C[0, 1] R given by (f ) = 0 f (t)dt, the
area under the curve y = f (x) (x [0, 1]). Since area is additive, is a
group homomorphism. Given a real number , if one consider the constant
function with constant value , then () = . Thus is surjective. The
kernel of consists of those continuous functions f in C[0, 1] for which the
area under the curve y = f (x) is 0. We see below by explicit construction
that the kernel of indeed is quite large.
Let f C[0, 1/2] be such that the graph of f lies entirely below the X-axis
and satisfies f (1/2) = 0. Consider the reflection of f along the line x = 1/2,
that is, the function g C[1/2, 1] given by
g(x) = f (1 x) (x [1/2, 1]).
Note that the area under f is same as the area under g. Now define h
C[0, 1] as follows:
h(x) = f (x) for x [0, 1/2]
= g(x) for x [1/2, 1].
Obviously, h C[0, 1] and (h) = 0.
Exercise 2.27 : Consider the map : C[0, 1] R given by (f ) = f (0).
Show that is a group homomorphism from (C[0, 1], +) onto (R, +).
Example 2.28 : Consider the subgroup GLn (R) of (Mn (R), ) of real n n
matrices with non-zero determinant. Also, consider the multiplicative group
(R , ) of real numbers. Consider the mapping det : GLn (R) R , which
sends an nn matrix to its determinant. Since determinant is multiplicative,
det is a group homomorphism. If R then determinant of the diagonal
matrix with diagonal entries , 1, , 1 is equal to . Thus det is surjective.
The kernel of consists of all matrices in GLn (R) whose determinant is 1.
The kernel of is commonly denoted by SLn (R).
A subgroup S of a group G is said to be normal if for every a S and
every b G, the conjugate bab1 is in S.
Trivially, any subgroup of an abelian group is normal.
Example 2.29 : The subgroup of rotations in Dn is normal in Dn . This
follows from the following facts:

NOTES ON GROUP THEORY

11

(1) Any two rotations commute.

(2) If r is a rotation and is a reflection then r 1 = r1
(see Example 2.11). Let us calculate the center of Dn . If follows from (2)
above that a rotation r commute with a reflection iff r2 = 1 iff r = r0 , r .
It follows that ZDn = {r0 , r } if n is even, and ZDn = {r0 } otherwise.
Example 2.30 : The subgroup of invertible diagonal matrices is not normal
in (GLn (R), ).
Proposition 2.31. The kernel of a homomorphism is a normal subgroup.
Proof. In the notations of Proposition 2.24,
(bab1 ) = (b)(a)(b1 ) = (b)e0 (b)1 = e0
in view of Remark 2.15(2).

Remark 2.32 : Note that SLn (R) is a normal subgroup of (GLn (R), ).
Example 2.33 : The center of (SLn (R), ) turns out to be the subgroup of
scalar matrices. To see this, let A SLn (R). If D SLn (R) is the diagonal
matrix with distinct diagonal entries then AD = DA forces that A must
be diagonal. By interchanging the role of A and D, one can see that the
diagonal entries A are identical.
Exercise 2.34 : Show that ZSLn (C) is isomorphic to (In , ) (see (2.3)).
Proposition 2.35. The center of a group is a normal subgroup.
Proof. Let a ZG and b G. For any g G, by associativity of G,
(bab1 )g = b(ab1 g) = b(b1 ga) = ga.
By a similar argument, g(bab1 ) = ag. Since a ZG , so does bab1 .

Let G be a group and let a G. If H is a subset of G, then the subset

aH = {ah : h H} of G is said to be a coset of H in G.
Remark 2.36 : Any two cosets contain same number of elements. In fact,
: aH H given by (ah) = h is a bijection.
Let H be a subgroup of G. Note that aH = bH iff a1 b H. If one
defines aH
= bH if a1 b then
= defines an equivalence relation on the
collection of cosets of H. Define G/H to be the collection of equivalence
classes corresponding to the relation
= . In particular, the cosets of H in G
are either disjoint or identical. In case G is finite, this statement becomes
very interesting.
Theorem 2.37. (Lagrange) Let H be a subgroup of a finite group G. Then
|G/H| = |G|/|H|.

12

NOTES ON GROUP THEORY

Proof. Note that G is the disjoint union of |G/H| number of (disjoint) cosets
of H. Since any coset contains |H| elements, |G| = |H| |G/H|.

Remark 2.38 : Let a G and let k be the smallest positive integer such
that ak = e (called the order of a). Then, since {e, a, , ak1 } is a subgroup
of G, order of a divides |G|. Thus the order of a is at most |G|.
Corollary 2.39. If |G| is a prime number then G is cyclic, that is, there
exists a G such that G = {e, a, , a|G|1 }.
Proof. Let a G \ {e}. Then the order of a divides |G|. Since |G| is prime,
the order of a is |G|, and hence G = {e, a, , a|G|1 }.

Exercise 2.40 : Let X be a finite set and let F be a collection of subsets of
X which is closed with respect to union and intersection. Show that there
exists an integer k such that |F| = 2k .
Hint. F endowed with the symmetric difference is a group.
If aH, bH G/H then we define aH bH = abH.
Proposition 2.41. (G/H, ) is a binary structure if H is normal in G.
Proof. We must check that abH is independent of representatives a and b
of aH and bH respectively. Suppose that aH = a0 H and bH = b0 H. Then
a1 a0 , b1 b0 H. A simple algebra shows that
(ab)1 a0 b0 = b1 a1 a0 b0 = b1 b0 (b01 a1 a0 b0 ).
It follows from the normality of H that (ab)1 a0 b0 G/H.

Remark 2.42 : If H is normal then the binary structure (G/H, ) is actually

a group with identity H. In particular, G/ZG is a group.
Proposition 2.43. Let : G G0 be a group homomorphism. Then
: G/ ker ran given by (a ker ) = (a) is a group isomorphism.
Proof. Note that is well-defined:
a ker = b ker iff ab1 ker iff (a)(b1 ) = e iff (a) = (b).
By Proposition 2.31 and Remark 2.42, G/ ker is a group. Since is a
homomorphism, so is . Clearly, is injective.

The following fact is analogous to rank-nullity theorem of Linear Algebra.
Corollary 2.44. Let : G G0 be a group homomorphism for a finite
group G. Then
|G| = | ker ||ran |.
Proof. By the last theorem, G/ ker is isomorphic to ran . In particular,
|G/ ker | = |ran |. The formula now follows from Lagranges Theorem. 
Example 2.45 : Consider the group Zn := Z/mod n with the binary operation addition modulo n. Define : Z6 Z6 by (k) = 2k. Then ker =
{0, 3} and ran = {0, 2, 4}. Obviously, we have |Z6 | = 6 = | ker ||ran |.

NOTES ON GROUP THEORY

13

3. Group Actions
Definition 3.1 : Let X be a set and G be a group. A group action of G on
X is a map : G X X given by (g, x) g x such that
(1) (gh) x = g (h x) for all g, h G and x X.
(2) e x = x for all x X.
If this happens then we say that G acts on X and X is a G-set.
Definition 3.2 : Let be a group action of G on X. For x X, let Gx
denote the subset {g x : g G} of X. Define an equivalence relation v on
X by setting x v y iff Gx = Gy. The equivalence class of x is known as the
orbit Ox of x.
Remark 3.3 : Note that
Ox = {y X : x v y} = {y X : y = g x for some g G} = Gx.
Note that X is the disjoint union of orbits of elements of X.
For x X, consider the function x : G Gx given by x (g) = g x.
Clearly, x is surjective. Note that x is bijective iff {g G : g x = x} =
{e}. This motivates the following definition.
Definition 3.4 : Let be a group action of G on X. For x X, the stabiliser
Sx of x is defined by {g G : g x = x}.
Remark 3.5 : Note that e Sx in view of (2) of Definition 3.1. Also, if
g, h Sx then (gh) x = g (h x) = g x = x by (1) of Definition 3.1.
Further, if g Sx then by the same argument g 1 x = g 1 (g x) = x.
Thus Sx is a subgroup of G.
In view of the discussion prior to the definition, |Ox | = |G| if |Sx | = 1.
We try to understand the notions of group action, orbit and stabiliser
through several examples.
Exercise 3.6 : Show that Rn acts on itself by translations: x y = x + y.
Find orbits and stabilisers of all points in Rn .
Example 3.7 : Consider the group (R , ) and the set Rn . Consider the
map (x1 , , xn ) := ( x1 , , xn ). By the associativity of R,
( ) (x1 , , xn ) = (( ) x1 , , ( ) xn ) = ( (x1 , , xn ))
for , R and x
= (x1 , , xn ) Rn . Clearly, 1 x
=x
for every x
Rn .

n
Thus R acts on R .
Let x
Rn . The orbit Ox is the punctured line in Rn passing through x

and the origin 0. Note that Sx = {1} of R if x

6= 0, and S0 = R .
Example 3.8 : Consider the group structure (T, ) and the unit ball B
centered at the origin. Then t z B for every t T and z B. Since

14

NOTES ON GROUP THEORY

multiplication is associative and 1 z = z, the unit circle T acts on B via the

complex multiplication.
The orbit Oz is the circle of radius |z| with origin as the center. The
stabiliser of all points except the origin is {1}. Clearly, S0 = T.
Exercise 3.9 : Find all subsets X of the complex plane such that the complex multiplication is a group action of (T, ) on X.
P
k l
Exercise 3.10 : Consider the power series f (z, w) =
k,l=0 ck,l z w in the
complex variables z and w. The domain of convergence Df of f is given by
{(z, w) C2 :

|ck,l ||z|k |w|l < }.

k,l=0

Show that the multiplicative group torus T T acts on Df via

(1 , 2 ) (z, w) = (1 z, 2 w).
Discuss orbits and stabilizers.
Let be a group action of G on X. We say that is transitive if there
is only one (disjoint) orbit in X, and that is free if every point in X has
trivial stabilizer.
Exercise 3.11 : Show that the symmetric group Sn acts on X = {1, , n}
via the action : (, j) (j). Show that is transitive. Find the
stabiliser of j.
Exercise 3.12 : Show that the symmetric group Sn acts on Rn via
(x1 , , xn ) = (x(1) , , x(n) ).
Describe orbits and stabilisers of x Rn such that x1 = 0, , xk = 0 for
some 1 k n.
Exercise 3.13 : For an n n matrix A and x
Rn , define by A x
= A
x.
For the group G and set X, verify that defines a group action:
(1) GLn (R) and Rn .
(2) On (R) (group of orthogonal matrices) and S (unit sphere in Rn ).
Find orbits and stabilisers of all points in X in both the cases.
Example 3.14 : Let us see an easy deduction of the following fact using
group action: S = {A = (aij ) GL2 (R) : ai1 + ai2 = 1 for i = 1, 2} is a
subgroup of GL2 (R). To see that, consider the group action of GL2 (R) on
R2 as discussed in the preceding exercise, and note that S is precisely the
stabilizer of the column vector (1, 1)T .
Example 3.15 : For a function f : R R and x R, consider the map
f x = f (x). Note that is not a group action of the group (C(R), +) on R.

NOTES ON GROUP THEORY

15

Exercise 3.16 : Consider the automorphism group A(D) of the unit disc D
endowed with the composition:
A(D) = {f : D D : f is a biholomorphism}.
Show that the action f z = f (z) of A(D) on D is transitive but not free.
Hint. Schwarz Lemma from Complex Analysis.
Exercise 3.17 : Let B denote the set of ordered bases (e1 , e2 ) of R2 . Show
that GL2 (R) acts B via A (e1 , e2 ) = (Ae1 , Ae2 ). Describe the orbit and
stabiliser of (e1 , e2 ).
Example 3.18 : Consider the group Z2 and the unit sphere Sn in Rn . Define
by 0 x = x and 1 x = x. Verify that is a free action of Z2 on Sn .
Note that Ox = {x, x} for any x Sn . Clearly, Sx = {0}. Although, we
do not required this fact, note that the real projective n-space RP n is the
space of orbits Ox endowed with the quotient topology.
The following example arises in the dynamics of projectile.
Exercise 3.19 : Verify that
t (x, y, z, v1 , v2 , v3 ) = (x + v1 t, y + v2 t, z gt2 /2 + v3 t, v1 , v2 , v3 gt)
is a group action of the additive group R on R6 , where g is a real constant.
Discuss orbits and stabilizers.
4. Fundamental Theorem of Group Actions
Theorem 4.1. Let G be a finite group acting on a set X. Then:
(1) If X is finite then there exist disjoint orbits Ox1 , , Oxk such that
|X| = |Ox1 | + + |Oxk |.
(2) The stabilizer Sx of x is a subgroup of G for every x X.
(3) (Orbit-Stabilizer Formula) For each x X,
|G/Sx | = |Ox | and |G| = |Ox ||Sx |.
(4) If y Ox then there exists h G such that Sx = hSy h1 .
(5) The map : G SX given by (g) = Mg is a group homomorphism,
where Mg SX is defined by Mg (x) = g x.
(6) ker() = xX Sx .
Proof. (1) This part follows from Remark 3.3.
(2) This is already noted in Remark 3.5.
(3) By (2) and the Lagranges Theorem, |G| = |G/Sx ||Sx |. Thus it suffices
to check that |G/Sx | = |Ox |. To see that, we define : G/Sx Ox by
(gSx ) = g x.
is well-defined and bijective: Note that gSx = hSx iff h1 g Sx iff
1
(h g) x = x iff g x = h x. Clearly, is surjective.

16

NOTES ON GROUP THEORY

(4) Let y Ox . Then x = h y for some h G, and hence by condition

(1) of Definition 3.1,
Sx = {g G : g x = x} = {g G : g (h y) = h y for some h G}
= {g G : h1 gh Sy } = hSy h1 .
(5) First note that is well-defined since Mg is bijective with inverse
Mg1 . Since Mgh = Mg Mh , is a group homomorphism.
(6) Note that
ker() = {g G : Mg = Me } = {g G : g x = x for all x X} = xX Sx .
This completes the proof of the theorem.

Corollary 4.2. Let be a group action of G on X. If is transitive then

|Sx | = |G|/|X| for every x X. If is free then |Ox | = |G| for every x X.
We will refer to as the permutation representation of G on X. We say
that is faithful if ker() = {e}.
Example 4.3 : Consider the square S with vertices v1 = (1, 1), v2 =
(1, 1), v3 = (1, 1), v4 = (1, 1). Consider the dihedral group D4 of symmetries of S and the set X = {(v1 , v3 ), (v2 , v4 )} of unordered pairs. Then
A (u, v) = (Au, Av) defines a group action of D4 on X.
Clearly, the orbit of any point is X, and hence is transitive. Also,
r S(v1 ,v3 ) S(v2 ,v4 ) . In particular, the action is not faithful.
If X is finite then SX is isomorphic to S|X| . Thus we have:
Corollary 4.4. Let be a permutation representation of G on X. If is
faithful then G is isomorphic to a subgroup of the permutation group S|X| .
Example 4.5 : Consider the group GL2 (Z2 ) of invertible matrices with
entries from the field Z2 . Then |GL2 (Z2 )| = 6. Consider the set X =
{e1 , e2 , e1 + e2 }, where e1 = [1 0]T and e2 = [0 1]T . Then GL2 (Z2 ) acts
on X via A x
= A
x.
Let denote the permutation representation of GL2 (Z2 ) on X. If A
ker then A
x=x
for every x
X. However, Aei is the ith column of A for
i = 1, 2. Therefore A is the identity matrix, and is faithful. By Corollary
4.4, GL2 (Z2 ) isomorphic to S3 .
Let p be a prime number. A group G is said to be a p-group if |G| = pk
for some positive integer k.
Corollary 4.6. Let G be a p-group acting on X. If
S = {x X : Ox is singleton},
then |S| = |X| mod p.
Proof. Let Ox1 , , Oxr be all disjoint orbits of X of size bigger than 1.
Then |X| = |S| + |Ox1 | + + |Oxr |. By Theorem 4.1, each |Oxi | divides
|G|, and hence a multiple of p. This gives |S| = |X| mod p.


NOTES ON GROUP THEORY

17

5. Applications
In this section, we discuss several applications of the fundamental theorem
of group actions to the group theory.
5.1. A Theorem of Lagrange.
Example 5.1 : The symmetric group Sn acts on the set C[z1 , , zn ] of
complex polynomials in n variables z1 , , zn via
p(z1 , , zn ) = p(z(1) , , z(n) ).
Let p be in C[z1 , , zn ]. Clearly, the identity permutation fixes p. Also,
( ) p(z1 , , zn ) = p(z (1) , , z (n) ) = p(z( (1)) , , z( (n) ))
= p(z (1) , , z (n) ) = ( p(z1 , , zn )).
The orbit of p is {p(z(1) , , z(n) ) C[z1 , , zn ] : Sn } and the
stabilizer of p is { Sn : p(z(1) , , z(n) ) = p(z1 , , zn )}.
Theorem 5.2. For any polynomial p C[z1 , , zn ], the number of different polynomials we obtain from p through permutations of the variables
z1 , , zn is a factor of n!.
5.2. A Counting Principle.
Example 5.3 : Let H, K be two subgroups of the group G. Then K acts
on the collection {aH : a G} of cosets of H by k (aH) := (ka)H.
The orbit of aH is {kaH : k K}. The stabiliser of aH is the subgroup
(a1 Ha) K.
Suppose H = K. Then H is normal in G iff the orbit of aH is singleton
for every a G.
Theorem 5.4. For subgroups H, K of a finite group G, let KH := {kh :
k K, h H}. Then
|KH| =

|K||H|
.
|K H|

Proof. Consider the group action K on the cosets of H as discussed in

Example 5.3. Then OH = {kH : k K} and SH = H K. By Theorem
4.1, |K| = |OH ||H K|. Also, since KH is the disjoint union of cosets in
OH and since |kH| = |H|,
|KH| = |OH ||H| =
This completes the proof of the corollary.

|K||H|
.
|H K|


18

NOTES ON GROUP THEORY

5.3. Cayleys Theorem.

Example 5.5 : Consider the group action of K on {aH : a G} as discussed in Example 5.3. The choice H = {e} and K = G gives the left
multiplication action of G on itself: (g, h) gh. Note that Og = G and
Sg = {e} for any g G. In particular, the permutation representation of G
on G is transitive and faithful.
Theorem 5.6. Every finite group is isomorphic to a subgroup of a symmetric group.
Proof. Consider the left multiplication action of G on itself. Let be the
corresponding permutation representation of G on X. By Corollary 4.4, it
suffices to check that is faithful. This is noted in the last example.

Exercise 5.7 : Let G be a group of order n. Prove:
(1) Let g1 , g2 , , G be such that g1 6= e and Hi ( Hi+1 , where Hi is
the subgroup generated by g1 , , gi . Then |Hi | 2i for each i.
(2) G can be generated by at most log2 n elements.
Use the Cayleys Theorem to conclude that the number of non-isomorphic
groups of order n does not exceed (n!)log2 n .
5.4. The Class Equation. Let us discuss another important group action
of G on itself.
Example 5.8 : Let G be a group. Define : G G G by g x = gxg 1 .
By Proposition 2.7 and associativity of G,
(gh) x = (gh)x(gh)1 = g(hxh1 )g 1 = g(h x)g 1 = g (h x)
for every g, h, x G. Also, since e x = exe1 = x for every x G, the map
defines a group action of G onto itself.
Note that Ox is the set of all conjugate elements of x. The stabiliser Sx
of x is the normalizer N (x) of x.
We will refer to the group action of the last example as the conjugate
group action of G. The orbit Ox will be referred as the conjugacy class of G.
Exercise 5.9 : Consider the conjugate group action of G on itself. Verify:
(1) If x, y ZG then x = g y for some g G iff x = y.
(2) Oa = {a} iff a ZG .
Theorem 5.10. Let G be a p-group. Then |ZG | is divisible by p.
Proof. Apply Corollary 4.6 to the conjugate group action of G on itself. 
Exercise 5.11 : Let G be a p-group such that |G| = p2 . Show that G is
abelian. Conclude that G is isomorphic either to Zp2 or Zp Zp .

NOTES ON GROUP THEORY

19

Theorem 5.12. Let G be a finite group. Then

X |G|
|G| = |ZG | +
,
|N (x)|
where this sum runs over one element x from each conjugacy class with
N (x) 6= G.
Proof. Consider the conjugate group action of G onto itself. We observed
in Example 5.8 that Sx = N (x). By Theorem 4.1, there exist disjoint orbits
of Ox1 , , Oxk such that
|G| = |Ox1 | + + |Oxk | =

|G|
|G|
|G|
|G|
+ +
=
+ +
.
|Sx1 |
|Sxk |
|N (x1 )|
|N (xk )|

Now if xi ZG then N (xi ) = G, and if xi , xj ZG for i 6= j then the conjugacy classes of xi and xj are disjoint (Exercise 5.9). The desired conclusion
follows immediately.

Remark 5.13
P : One may derive Theorem 5.10 from the class equation:
|ZG | = |G| |G|/|N (x)|, where N (x) is a proper subgroup of G.
Example 5.14 : Consider the conjugate group action of the dihedral group
D4 on itself. We already noted in Example 2.29 that ZD4 = {r0 , r }. In
particular, N (r0 ) = D4 = N (r ).
Note that the normalizer of any rotation x contains exactly 4 elements
(all the rotations in D4 ). Note further that the normalizer of any reflection
x contains exactly 4 elements (2 elements in ZD4 , x = and + ). Note
also that r/2 and r3/2 are conjugate to each other, and hence belong to
same conjugacy class. Note next that /2 and (resp. 3/2 and 2 ) are
conjugates. Hence the class equation for D4 is
(|D4 | = 8) = (|ZD4 | = 2) + (|Or/2 | = 2) + (|O/2 | = 2) + (|O3/2 | = 2).
Exercise 5.15 : Verify that the conjugacy classes of the permutation group
S3 are {(1)}, {(1, 2), (1, 3), (2, 3)}, {(1, 2, 3), (1, 3, 2)}. Conclude that the class
equation of S3 is
(|S3 | = 6) = (|ZS3 | = 1) + (|O(1,2) | = 3) + (|O(1,2,3) | = 2).
5.5. Cauchys Theorem.
Example 5.16 : Let G be a group and let H be the multiplicative group
{1, 1}. Consider the action 1 g = g and (1) g = g 1 of H on G. It is
easy to verify that is a group action.
Let g G. The orbit of g is {g, g 1 } if g 6= e, and {e} otherwise. The
stabiliser of g is {1} if g 2 6= e, and {1, 1} otherwise.
Here is a particular case of the Cauchys Theorem.
Exercise 5.17 : Prove that there exist odd number of elements of order 2
in a group of even order.

20

NOTES ON GROUP THEORY

Hint. Apply Theorem 4.1 to the action discussed in Example 5.16.

Example 5.18 : Let G be a group and let p be a prime number. Let H
denote the group generated by the permutation Sp given by (j) = j +1
for j = 1, , p 1 and (p) = 1. Note that |H| = p. Consider the set
X = {(g1 , , gp ) : g1 , , gp G, g1 gp = e}.
Since there are p parameters and 1 equation, p 1 variables are free, and
hence |X| = |G|p1 .
Suppose (g1 , , gp ) X. Then g1 commutes with g2 gp , and hence
(g(1) , , g(p) ) X. By finite induction, we get (gj (1) , , gj (p) ) X
for j = 1, , p. This enables us to define
(g1 , , gp ) = (g (1) , , g (p) ) ( H, (g1 , , gp ) X).
Then is indeed a group action of H on X. We verify only condition (2) of
Definition 3.1:
( ) (g1 , , gp ) = (g (1) , , g (p) ) = (g( (1)) , , g( (p) ))
= (g (1) , , g (p) ) = ( (g1 , , gp )).
The orbit of (e, , e) is {(e, , e)}. More generally, |O(g1 , ,gp ) | = 1 iff
g1 = = gp and g1p = e. Finally, the stabiliser of any element in X consists
only the identity permutation in Sp .
Theorem 5.19. Let G be a finite group and p be a prime such that p divides
|G|. If P denotes the set of elements of G of order p, then |P | 1 mod p.
Proof. Consider the action of H on X as discussed in Example 5.18. Recall
that |H| = p, |X| = |G|p1 , and the fact that |O(g1 , ,gp ) | = 1 iff g1 = g2 =
= gp and g1p = e.
By Theorem 4.1, there exist disjoint orbits of Ox1 , , Oxk such that
X
|X| = |Ox1 | + + |Oxk | = 1 +
|Oxi |.
Oxi 6=O(e, ,e)

Since |X| = |G|p1 and G is a p-group, 1 + xi 6=(e, ,e) |Oxi | = lp for some
positive integer l. Again, by Theorem 4.1, |Oxi | divides |H| = p, and hence
|Oxi | is either 1 or p. Let Y = {xi : Oxi 6= O(e, ,e) , |Oxi | = 1}. It follows
that 1 + |Y | + (k 1 |Y |)p = lp. Thus we have |Y | = 1 mod p.
Define : P Y by (g) = (g, , g). Clearly, is injective. If
(g1 , , gp ) Y then |O(g1 , ,gp ) | = 1. Then we must have g1 = = gp and
g1p = e. Thus (g1 ) = (g1 , , gp ), and hence is surjective.

P

Remark 5.20 : A group of order 6 must contain an element of order 3.

In particular, there exists no group of order 6 containing identity and 5
elements of order 2.
Example 5.21 : Let G be a group of order 6. Then G contains an element
x of order 3 and an element y of order 2. If i is not a multiple of 3 then

NOTES ON GROUP THEORY

21

(xi )3 = (x3 )i = e and (xi )2 6= e. Similarly, the order of y j is 3 if j is not

a multiple of 2. Now if xi y j = xr y s then xir = y sj , and hence i = r
mod 3 and s = j mod 2. Thus G = {xi y j : 0 i 2, 0 j 1}. Now
yx G, and clearly, yx
/ {e, x, y, x2 }. Hence there are only two choices of
2
yx, namely, xy or x y. If xy = yx then G is the cyclic group of order 6. If
yx = x2 y or yxy 1 = x1 then G is isomorphic to the dihedral group D3 .
5.6. First Sylow Theorem.
Example 5.22 : Let G be a group. For a positive integer n |G|, let Fn
denote the collection of all subsets A of G such that |A| = n. Note that


|Fn | = |G|
n . Then G acts on Fn by (g, A) gA. Indeed, |gA| = |A|,
eA = A and (g1 g2 ) A = g1 (g2 A).
The orbit OA of A equals {gA : g G} and the stabilizer SA of A equals
{g G : gA = A}. If a A then SA a = {ga : gA = A} A.
Suppose |G| = pn q, where p is a prime not dividing q. Then a Sylow psubgroup is a subgroup of order pn . A subgroup is called p-subgroup if it is a
p-group.
Remark 5.23 : Let a belong to a Sylow p-subgroup such that a 6= e. By
the Lagranges Theorem, the order a is pr for some positive integer r. Then
r1
the order of ap
is precisely p.
The First Sylow Theorem partly generalizes the Cauchys Theorem.
Theorem 5.24. Let G be a finite group. If p is a prime divisor of |G| then
G has a Sylow p-subgroup.
Proof. Suppose |G| = pn q, where q is not divisible by p. Consider the group
action of G on X := Fpn as discussed in Example 5.22. We need the fact
n

that |Fpn | = ppnq is not divisible by p. Then by Theorem 4.1(1), there exists
A Fpn such that |OA | is not divisible by p. By the orbit-stabilizer formula,
pn q = |SA ||OA |, and hence |SA | is divisible by pn . Also, |SA | = |SA a| and
SA a A for any a A. It follows that |SA | = pn .

5.7. Second Sylow Theorem.
Remark 5.25 : Since |aHa1 | = |H|, if H is a Sylow p-subgroup of G then
so is the conjugate aHa1 of H. If G has only one Sylow p-subgroup H then
H is necessarily normal in G.
Theorem 5.26. Let G be a finite group and let p be a prime divisor of |G|.
If H is a Sylow p-subgroup of G and K is a p-subgroup of G then there exists
a G such that K aHa1 .
In particular, any two Sylow p-subgroups are conjugate.
Proof. Consider the group action of K on the collection X = {aH : a G} of
cosets of H by k (aH) := (ka)H (see Example 5.3). Let S denote the set of

22

NOTES ON GROUP THEORY

cosets aH with single-ton orbits. Since K is a p-subgroup, by Corollary 4.6,

|S| = |X| mod p. However, by the Lagranges Theorem, |X| = |G|/|H| = q,
where q is not divisible by p. This implies that |S| = q mod p, and hence
S is non-empty. Let aH S. Then kaH = aH for every k K, that is,
a1 ka H for every k K. This completes the proof of the first part.
If in addition K is also a p-Sylow subgroup then |K| = pn = |H| =
|aHa1 |. By the first part, we must have K = aHa1 in this case.

Remark 5.27 : Let H be a Sylow p-subgroup of G. Then H is a unique
Sylow p-subgroup of G iff H is normal in G.
5.8. Third Sylow Theorem.
Example 5.28 : Let G be a group and H, K be subgroups of G. Let X :=
{aHa1 : a G} be a collection of subgroups of G. Then K acts on X by
(g, L) gLg 1 . We just check condition (2) of Definition 3.1:
(gh) L = (gh)L(gh)1 = g(hLh1 )g 1 = g (h L)
for any g, h K and L X.
Suppose K := G. Then the orbit of any element in X is the entire X.
Also, the stabiliser of K X is the subgroup {g G : gK = Kg}, the
normalizer N (K) of K.
Theorem 5.29. Let G be a group such that |G| = pn q, where p is a prime
number which does not divide q. Then the number Np of Sylow p-subgroups
of G divides q. Moreover, Np = 1 + kp for some non-negative integer k.
Proof. Let H be a Sylow p-subgroup of G. Consider the group action of
K := G on X as discussed in Example 5.28. By the Second Sylow Theorem,
X consists of all Sylow p-subgroups of G. Thus Np = |X|. Recall that OH =
X and SH = N (H). By the orbit-stabilizer formula, |G| = |X||N (H)|. In
particular, Np divides |G|. Since H N (H), by the Lagranges Theorem,
|H| divides |N (H)|. It follows that Np = |G|/|N (H)| divides |G|/|H| = q.
In particular, Np is not divisible by p.
Consider the group action of K := H on X as described in Example 5.28.
Let S = {L X : OL = {L}}. By Corollary 4.6, |S| = |X| mod p. Since p
does not divide Np = |X|, S is non-empty. Thus there exists L X such
that OL = {L}. It follows that hL = Lh for every h H, that is, H N (L).
Thus H and L are subgroups of N (L). Also, since |N (L)| = |G|/Np = pn q 0
for some divisor q 0 of q, H and L are indeed Sylow p-subgroups of N (L).
However, since L is normal in N (L), by the Second Sylow Theorem, there
is only one Sylow p-subgroup of N (L). Thus H = L. This shows that S has
only one element. In particular, 1 = |X| mod p as desired.

A group G is called simple if it has no normal subgroup.
Example 5.30 : Let G be a group of order 12. The possible choices for N2
are 1 and 3. The possible choices for N3 are 1 and 4. If N3 = 4 then G must

NOTES ON GROUP THEORY

23

have 8 elements of order 3. Indeed, any two Sylow 3-subgroups intersects

trivially in view of the Lagranges Theorem. In case N3 = 4, there can be
only one Sylow 2-group of order 4. In any case, G is not simple.
Exercise 5.31 : Show a group of order 40 has a normal, Sylow 5-subgroup.
Example 5.32 : Consider the group GL2 (Zp ), where Zp is the multiplicative
group {0, 1, , p 1} with binary operation multiplication modulo p for
a prime number p. Any element in GL2 (Zp ) is obviously determined by 4
elements in Zp , out of which a column can be chosen in p2 1 ways, and
then the remaining column should not be a Zp -multiple of the first column
chosen in p2 p ways. Thus |GL2 (Zp )| = p(p 1)2 (p + 1). By the Sylow
Third Theorem, the number Np of Sylow p-subgroups is either 1 or p + 1.
Produce two Sylow p-subgroups of GL2 (Zp ) to conclude that Np = p + 1.
Corollary 5.33. Let Gpq be a group of order pq, where p and q are prime
numbers such that q < p. Then:
(1) Gpq has only one Sylow p-subgroup Hp .
(2) If p 6= 1 mod q then Gpq has only one Sylow q-subgroup Hq .
(3) If Gpq has only one Sylow q-subgroup Hq then Gpq is cyclic.
(4) If p 6= 1 mod q then Gpq is cyclic.
(5) Gp2 is either abelian or isomorphic to the dihedral group Dp .
Proof. (1) By the Third Sylow Theorem, Np divides q and Np = 1 mod p.
Thus Np q < p, and hence Np = 1.
(2) Again, by the Third Sylow Theorem, Nq divides p and Nq = 1 mod q.
Either Nq is p or 1. If p 6= 1 mod q then Nq must be 1.
(3) Suppose Gpq has only one Sylow q-subgroup Hq . By Remark 5.27,
Hp and Hq are normal in Gpq . Clearly, Hp and Hq are cyclic of order p
and q respectively. By the Lagranges Theorem, Hp Hq is trivial. Let x
and y denote the generators of Hp and Hq respectively. Since Hp is normal
in Gpq , (xy)(yx)1 = x(yx1 y 1 ) Hp . Also, since Hq is normal in Gpq ,
(xy)(yx)1 = (xyx1 )y 1 Hq . Since Hp Hq = {e}, xy = yx. It is easy to
see that the order of xy is pq. In particular, Gpq is a cyclic group generated
by xy.
(4) This follows from (2) and (3).
(5) Let x and y denote the generators of Hp and H2 respectively. Thus
p
x = e and y 2 = e. Since Hp is normal, yxy = yxy 1 Hp . Thus yxy = xj
2
for some 0 j < p. But then x = y 2 xy 2 = yxj y = xj , and hence j 2 = 1
mod p. The only possible choices of j are 1. If j = 1 then Gp2 is abelian. If
i = 1 then the relations xp = e, y 2 = e, yxy = x1 determines the dihedral
group Dp .

Remark 5.34 : Every group of order 15 is cyclic.

24

NOTES ON GROUP THEORY

Exercise 5.35 : Consider the group G7.3 . Verify the following:

(1) G7.3 has unique Sylow 7-subgroup, and G7.3 has k Sylow 3-subgroups
H3,j (j = 1, , k), where possible values of k are 1 and 7.
(2) Suppose H7 is generated by x and H3,1 is generated by y. There
exists positive integer i < 7 such that yx = xi y. Further, i satisfies
i3 = 1 mod 7, that is, i = 1, 2, 4.
(3) Let Gi denote the group generated by x, y satisfying x7 = e, y 3 = e,
and yx = xi y. Then G1 is abelian and isomorphic to H7 H3,1 .
(4) Define : G2 G4 by (x) = x and (y) = y 2 . Show that extends
to an isomorphism.
Conclude that there are two isomorphism classes of groups of order 21.
Corollary 5.36. Let p, q be primes. Then every group G of order p2 q is
not simple.
Proof. Let Hp , Hq denote Sylow p-subgroup and Sylow q-subgroup of G
respectively.
Suppose q 6= 1 mod p. By the Third Sylow Theorem, Hp is the only
Sylow p-subgroup, and hence normal in G.
Suppose p2 6= 1 mod q. Then p 6= 1 mod q. By similar reasoning, Hq is
a normal Sylow q-subgroup of G.
Suppose q = 1 mod p and p2 = 1 mod q. This implies q > p. But then
q must divide p + 1 and p divides q 1. This is possible iff p = 2 and q = 3.
The desired conclusion follows from Example 5.30.

Exercise 5.37 : Let G be a group of order 56 and let Np denote the number
of Sylow p-subgroups of G. Verify the following:
(1) Either N7 = 1 or N7 = 8.
(2) If N7 = 8 then G contains 48 elements of order 7.
(3) Either N7 = 1 or N2 = 1.
(4) G is not simple.
Exercise 5.38 : Show that a finite group with every normal and abelian
Sylow subgroup is necessarily abelian.
6. Structure Theorem for Finite Abelian Groups
We will always assume that an abelian group G carries addition as the
binary operation. We say that G is the direct sum H1 H1 Hk of
subgroups H1 , H2 , , Hk of G if G = H1 +H2 + +Hk and Hi Hj = {0}
for all i 6= j.
The most basic example of finite abelian groups is the cyclic group Cn of
order n. It turns out that this forms a building block in the representation
theorem for abelian groups.
Exercise 6.1 : Let m, n N be coprime. Show that there exist x, y Cmn
of order m and n respectively such that Cmn =< x > < y >, where

NOTES ON GROUP THEORY

25

< a > denotes the cyclic group generated by a Cmn . Conclude that Cmn
is isomorphic to Cm Cn .
Remark 6.2 : Any cyclic group is isomorphic to the direct sum of finitely
Q
k
many cyclic groups of prime-power order. In fact, if m = lj=1 pj j , where
pj are distinct primes and kj are positive integers then Cm is isomorphic to
the direct sum lj=1 C kj .
pj

To understand finite abelian groups, in view of the last remark, it suffices

to understand the structure of non-cyclic abelian groups.
For an n n matrix A, the ith row is denoted by Ai .
Exercise 6.3 : For n 2, let a1 , , an be integers with gcd 1. For n 2,
consider the statement Pn : Then there exists A SLn (Z) such that A1 =
[a1 an ]. Prove Pn by induction on n by verifying:
(1) P2 holds true.
(2) Assume Pn1 . Let d be the gcd of a1 , , an1 . Then there exists
B SLn1 (Z) such that B1 = [b1 bn1 ], where bi = ai d1 .
(3) Let B be as ensured by (2). Choose s, t Z such that san + td = 1.
Let A be such that A1 = [a1 an ], Ai = [Bi 0] (i = 2, , n 1),
and An = [c1 cn1 t], where ci = ()n sbi . Verify that A SLn (Z).
Lemma 6.4. Let G be an abelian group such that x1 , , xn G are generators of G. If X = [x1 xn ]t and A SLn (Z) then the entries of AX
are generators of G. In particular, if y1 = a1 x1 + + an xn , then there exist
y2 , , yn such that y1 , , yn G are generators of G.
Proof. Note that if A SLn (Z) then A1 SLn (Z). If x G then there
exists k1 , , kn Z such that x = k1 x1 + kn xn . Now if AX = Y then
x = k1 (A1 X)1 + + kn (A1 X)n is a Z-linear combination of y1 , , yn .
This completes the proof of the first part. To see the remaining part, let
A SLn (Z) with first row [a1 an ] as ensured by the last exercise, and
take yi = (A1 X)i for i = 2, , n.

The last lemma may be interpreted as:
Proposition 6.5. Let G be a finitely generated abelian group. Let F = {F :
F is a set of generators of G}. Let n = minF F |F |. Let Fn = {[g1 gn ]T :
{g1 , , gn } F }. Then the group SLn (Z) acts on Fn via A F = AF.
Theorem 6.6. If G is a finitely generated abelian group then G is the direct
sum of cyclic groups.
Proof. Let {x1 , , xn } Fn be such that xn has the minimal order k,
where Fn is as defined in the last remark. We prove by induction that
G = Zx1 Zxn . The case n = 1 is trivial. Let H be the proper
subgroup of G generated by x1 , , xn1 . By the induction hypothesis, H =
Zx1 Zxn1 . Thus it suffices to check that G = H Zxn .

26

NOTES ON GROUP THEORY

Clearly, G = H + Zxn . Suppose H Zxn 6= {0}. Then there there exist

a1 , , an Z such that a1 x1 + + an1 xn1 = an xn 6= 0., In particular,
an
an < k. If gcd of a1 , , an is d then y1 = ad1 x1 + + an1
d xn1 d xn
satisfies dy1 = 0. Also, by the preceding proposition, there exist y2 , , yn
G such that {y2 , , yn , y1 } Fn . But then the order l of y1 must be less
than or equal to the order k of xn , that is l k. Also, since dy1 = 0, l
divides d an < k. This is not possible.

Remark 6.7 : Any finite abelian group is isomorphic to the direct sum of
finitely many cyclic groups of prime-power order. In particular, the abelian
group of order 16, up to isomorphism, are
C16 , C2 C8 , C4 C4 , C2 C2 C4 , C2 C2 C2 C2
(see Remark 6.2).
Exercise 6.8 : List all abelian groups of order 216 up to isomorphism.
Acknowledgment. I am thankful to all the students of MTH 751 for their
active participation and several valuable inputs.
References
[1] M. Artin, Algebra, Eastern Economy Edition, 1996.
[2] J. Clay, The Punctured Plane is Isomorphic to the Unit Circle, Journal of Number
Theory, 1, 500-501, 1969.
[3] K. Conrad, Group Actions, available online.
[4] S. Kumaresan, Group Actions; Sylow Theorems: Expository Articles (Level 1), private communication.
[5] S. Kumaresan, Basis Theorem for Finitely Generated Abelian Groups: Expository
Articles (Level 2), private communication.
[6] P. Morandi, Group Actions, available online.