Algebraic Structures
Algebraic Structures
RALPH STOHR
1. Binary Operations
Definition. A binary operation on a non-empty set S is a rule that assigns to
each ordered pair of elements of elements of S a uniquely determined element of S.
The element assigned to the ordered pair (a, b) with a, b S is denoted by a b.
Remark. In other words, a binary operarion of a set S is a function : SS S
from the Cartesian product S S to the set S. The only dierence is that the value
of the function at an ordered pair (a, b) is denoted by a b rather than ((a, b)).
Examples. Let S = N = {1, 2, 3, . . .}.
(1) a b = max(a, b)
e.g. 2 3 = 3, 3 2 = 3, 3 3 = 3.
(2) a b = a
e.g. 2 3 = 2, 3 2 = 3, 3 3 = 3.
(3) azb = ab
e.g. 2z3 = 23 = 8, 3z2 = 32 = 9, 3z3 = 33 = 27.
Definition. A binary operation on a set S is commutative, if
ab=ba
a, b, S.
The binary operation is commutative, but the binary operations and z are
not commutative.
Let be a binary operation on a set S. and let a, b, c S. Consider the
expression
a b c.
This expression doesnt have a meaning since gives only a meaning to ordered
pairs of elements. In fact, there are two ways of making a b c respectable, namely
(a b) c
and
a (b c).
34=
3 (2 4) =
34=
4.
RALPH STOHR
34=
3 (2 4) =
32=
3.
= (33 )z3 =
3z(3z3) = 3z(33 ) =
(33 )3 = 39
327 .
So, in general,
(azb)zc = az(bzc).
Definition. A binary operation on a set S is called associative, if
(a b) c = a (b c).
a, b, c S.
In our examples,
is both commutative and associative,
is not commutative, but associative,
z is neither commutative nor associative.
If is an associative operation on S, then we can write a b c for the common
value of (a b) c and a (b c):
a b c = (a b) c = a (b c).
NB. This works only for associative operations!
Our three examples , , z are of course articially made up operations. But
there are many natural examples of binary operations.
(a) = +.
This is also a binary operation on the set Sn , the set of all permutations of the set
= {1, 2, . . . , n}.
(f) Addition of vectors in a vector space is a binary operation. (NB. Scalar
multiplication is not a binary operation.)
ALL the binary operations in Examples a) - f) are associative, and all EXCEPT
matrix multiplication and composition of functions are commutative.
3 important points about binary operations:
(i) The result of the operation must be an element of S. This fails, for example,
for + on the set S = {1, 0, 1} (as 1 + 1 = 2
/ S).
(ii) The operation must be dened for all elements of S. This fails, for example
for A B = A1 BA on Mn (R) (as the matrix A1 may not exist).
(iii) The result of the operation must be uniquely determined. This fails, for
example, if we set
ab=c
c2 = ab
where
on
a1
ai
an
a1
a1 a1
an a1
a2
a1 a2
an a2
aj
ai aj
an
a1 an
an an
0
1
2
0
0
0
0
1
0
1
2
2
0
2
1
RALPH STOHR
IDENTITY ELEMENTS
Definition. Let be a binary operation on a set S. An element e S is an
identity element for if
ea=ae=a
a S.
a)
a)
(c) = + on Zn : e = 0.
= on Zn : e = 1.
(d) The identity element for addition of n n matrices is e = On (the zero
matrix).
The identity element for matrix multiplication on Mn (R) is e = In (the identity
matrix).
(e) = on F(): e = id (the identity map dened by id(x) = x for all x ).
Indeed, for any function : we have id = id = .
(f) Addition of vectors in a vector space: e = 0, the zero vector.
Fact: If there is an identity element for a binary operation, then this element is
unique.
Proof. Suppose e and f are identity elements for a binary operation on a set
S. Then e f = e (since f is an identity element). At the same time, e f = f
(since e is an identity element). Hence e = e f = f , and therefore e = f .
ASSOCIATIVE POWERS
For an associative binary operation on a set S, and a natural number n we
dene
an = a
| a {z a} .
n
= (f f f )(x)
= f (f (f (x)))
=
2(2(2x + 3) + 3) + 3
8x + 21.
am an = am+n
(ii)
(am )n = amn .
Proof.
am an
= |a a {z a} a
| a {z a}
m
= a
| a {z a}
m+n
= a
(am )n
m+n
= |a a {z a} a
| a {z a}
m
m
|
{z
}
= |a a {z a}
mn
= amn .
RALPH STOHR
2. Groups
Definition. Let G be a non-empty set with a binary operation dened on it.
The system (G, ) is a group if the following three conditions are satised:
(G1) is associative.
(G2) There exists an element e G such that
eg =ge=g
g G
n
=
(1) (2) (n)
is
(1)
1
(2)
2
(n)
n
)
.
RALPH STOHR
where p is a prime.
Note: The set Mn (R) of all n n-matrices over R is not a group under matrix
multiplication (since it contains matrices which are not invertible). However, Mn (R)
is a group under matrix addition.
ELEMENTARY PROPERTIES OF GROUPS
Lemma 2.2 (The Cancelation Lemma) Let (G, ) be a group, and let a, b, c G.
Then
(i)
(ii)
ab=ac
ba=ca
implies
implies
b = c,
b = c.
For example,
|Z| = ,
|R | = ,
|Zn | = n,
|Sn | = n!,
|GL(n, R)| = .
Answer: pn .
Return to the original question: What is the order of GL(n, Zp ), p a prime? In
other words, what is the number of invertible n n matrices with entries in the
ZP ?
Recall from Linear Algebra: A matrix A Mn (Zp ) is invertible if and only if
its column vectors are linearly independent. So we need to count the number of
n n matrices with linearly independent columns. In an invertible matrix, the rst
column can be any column except the zero column. Hence the number of possible
choices for the rst column is pn 1. Once the rst column is chosen, the second
column can be any vector that is not in the span of the rst column. There are
p vectors in that span, and hence there are pn p possible choices for the second
column. Once the rst and second column are chosen, the third column can be any
vector that is not in the span of the two rst column vectors. That span contains p2
vectors. Hence there are pn p2 possible choices for the third column. Continuing
in this way we see that the number of choices for the (k + 1)-st column, once the
rst k have been chosen, is pn pk . Hence the total number of invertible matrices
in Mn (Zp ) is
|GL(n, Zp )| = (pn 1)(pn p)(pn p2 ) (pn pn1 ).
In particular,
|GL(2, Z2 )| = 3 2 = 6,
|GL(2, Z3 )| = 8 6 = 48,
|GL(3, Z2 )| = 7 6 4 = 168.
Definition. A group G is called Abelian (or commutative), if the binary operation of G is commutative, i.e. if gh = hg for all g, h G.
For example,
all groups of numbers are abelian,
Zn is abelian,
Sn is not abelian for n 3, since
(
)
(
1 2 3
1
= (12)(23) = (23)(12) =
2 3 1
3
2
1
3
2
RALPH STOHR
10
GL(n, R) is not abelian for n 2, and GL(n, Zp ) is not abelian for all
n 2. For example,
(
)(
) (
)(
)
0 1
1 1
1 1
0 1
=
.
1 0
0 1
0 1
1 0
Indeed,
(
LHS =
0
1
1
1
(
RHS =
1 1
1 0
)
.
POWERS IN A GROUP
Recall that for associative binary operations we have dened the notion of powers
for exponents that are positive integers. In group notation this is an = aa
a} for
| {z
n
for n > 0;
a ,
n
a =
e,
for n = 0;
1 |n|
(a ) , for n < 0.
With this denition the usual power rules remain valid, i.e. we have for any element
of a group G and any integers m, n
am an = am+n
(am )n = amn .
and
For example,
a5 a3
= aaaaaa1 a1 a1
1 1 1
= aaaa aa
| {z } a a
=e
= aaaaa1 a1
= aaa |aa{z1} a1
=e
1
= aa aa
| {z }
=e
= aa
= a2
Finally, for multiplication tables of groups we have an important special property.
Lemma 2.4. Let G = {a1 , a2 , . . . , an } be a nite group. Then in any row and
column of the multiplication table of G, each element of G appears exactly once.
Proof. We assume that the elements a1 , a2 , . . . , an of G are distinct as listed.
The i-th row of the multiplication table is
ai a1 , ai a2 , . . . , ai an .
()
11
3. Subgroups
Let G be a group and let H be a subset of G. Then, for any two elements
g, h H, the binary operation of G gives a meaning to the product gh. This will
be an element of G, which may or may not be in H. If gh H for all g, h H, we
say that H is closed under the binary operation of G.
For example, consider Z, the additive group of integers ((Z, +) in our old notation), and its subsets
H1
= {. . . , 4, 2, 0, 2, 4, 6, . . .} Z
H2
= {1, 2, 3, 4, 5, 6, . . .} Z
H3
= {. . . , 3, 1, 1, 3, 5, 7, . . .} Z
i.e. the even integers, the positive integers and the odd integers. H1 and H2 are
closed under the operation +, but H3 is not closed, (e.g. 1 + 1 = 2
/ H3 ). If a
subset H is closed under the binary operation of G, then the group operation of G
gives us a binary operation on the subset H. The group operation of G induces a
binary operation of H.
Q: IF H is closed under the binary operation of G, is H a group with respect to
the induced operation?
For example, the even integers H1 are a group (0 is the identity element and for
each even integer 2k, the element 2k is the (additive) inverse). But the positive
integers H2 do not form a group (e.g. because they do not contain an identity
element).
Definition. A non-empty subset H of a group G is a subgroup, if H is closed
under the binary operation of G and H is itself a group under the induced operation.
Notation: H G.
In detail: A subset H G (H = ) is a subgroup if
gh H. (closure)
(1) g, h H
(2) g, h, k H
(gh)k = g(hk).
(3) e H : e h = he = h h H.
RALPH STOHR
12
(4) h H h H :
hh = h h = e .
Condition 1 is the closure condition and Conditions 2-4 are the normal group
axioms.
Comments on conditions 1-4:
(a) Condition 1 is essential. Without it we do not have a binary operation on
H, and then the question of whether or not H is a group doesnt make sense.
(b) The Condition 2 holds automatically (because it holds for all g, h, k G,
and hence, in particular, for all g, h, k H).
(c) If H has an identity element e , then e = e, i.e. the identity element of H
coincides with the identity element of G.
Proof. ee = e since e is the identity element of G. But also e e = e (since e
is the identity element of H and e H). Hence ee = e e in G. Hence e = e by
the Cancelation Lemma, applied in G.
Therefore, Condition 3 is equivalent to the condition e H.
(d) Now Condition 4 reads as follows: h H h H : hh = h h = e, and this
is equivalent to the condition h H : h1 H.
After these comments it is not hard to prove the following Theorem which provides an easy criterion for deciding whether of not a given subset of a group is a
subgroup.
Theorem. (The Subgroup Criterion) Let G be a group. A non-empty subset
H of G is a subgroup if and only if the following two conditions hold:
(i) g, h H :
(ii) h H :
gh H.
h H.
1
13
e
e
e
(123) (123)
(132) (132)
(123) (132)
(123) (132)
(132)
e
e
(123)
we see, that this set is closed under composition of permutations, and that each
element has an inverse.
{(
)
}
1 n
(e) H =
: n Z GL(2, R)
0 1
Indeed, we have
1
0
n
1
(
)(
1 n
0 1
1
0
m
1
(
=
)1
(
=
1 n+m
0
1
1 n
0 1
)
,
)
,
RALPH STOHR
14
so (ii) is okay.
Now some theory:
Lemma 3.1. Let G be a group, H G and K G. Then H K G.
Proof. We have that e H and e K, so e H K, so H K = , and hence
we can apply the Subgroup Criterion to prove that H K is a subgroup.
To check Condition (i), suppose that g, h H K. Then g H and h H,
and hence gh H (by Condition (i) of the Subgroup Criterion since H G).
Likewise, g, h H K means that g K and h K, and hence gh K (again
by Condition (i) of the Subgroup Criterion since K G). Consequently, gh H
and gh K, and hence gh H K. So Condition (i) of the Subgroup Criterion
holds. Now, if g H K, then g H and hence g 1 H (by Cond. (ii) of the
Subgroup Criterion). Likewise, g K and hence g 1 K (again by Cond. (ii)
of the Subgroup Criterion). Hence g 1 H and g 1 K, so g 1 is in H K,
and this means that Cond. (ii) of the Subgroup Criterion is satised. The Lemma
follows.
Note: In the proof we have used the Subgroup Criterion in both directions.
Remark: If H, K G, then H K is not necessarily a subgroup of G. For
example,
2Z = {2k | k Z} Z
and
3Z = {3k | k Z} Z,
but
2Z 3Z Z
since, for example, 2 + 3 = 5
/ 2Z 3Z.
CYCLIC SUBGROUPS
For a group G and an element a G, let
a = {ak | k Z} = {. . . , a2 , a1 , a0 = e, a, a2 , a3 , . . .}.
denote the set of all powers of a in G.
Examples: (a) In R we have
1 1 1
2 = {(2)k | k Z} = {. . . , , , 1, 2, 4, 8, 16, . . .}
8 4 2
and
1 = {(1)k | k Z} = {1, 1}.
15
= (123)
(123)2
= (123)(123) = (132)
(123)
= (123)3m+r
= (123)3m (123)r
= ((123)3 )m (123)r
= em (123)r
= (123)r .
Hence
(123), if k 1
k
(123) =
(132), if k 2
e,
if k 0
mod 3;
mod 3;
mod 3.
Consequently,
(123) = {e, (123), (132)}.
Lemma 3.2.. Let G be a group, a G. Then a is a
subgroup of G. Moreover, a is abelian.
Proof. We use the Subgroup Criterion. Clearly, a = since a a. Now, if
x, y a, then x = ak and y = am for some integers k, m. But then
xy = ak am = ak+m a.
Also
x1 = (ak )1 = ak a.
Hence a satises the two conditions of the Subgroup Criterion, so it is a subgroup.
Finally, ak am = ak+m = am+k = am ak , i.e. any two powers of a commute.
Hence a is abelian.
a is called the cyclic subgroup generated by a in G.
THE ORDER OF AN ELEMENT
Definition. Let G be a group, a G. The smallest natural number m such
that am = e is called the order of a. If there is no such natural number, we say
that a has infinite order.
Examples:
(a) In R , 2 has innite order, 1 has order 2.
(b) In S3 , (123) has order 3.
RALPH STOHR
16
a contradiction.
17
RALPH STOHR
18
Examples:
(a) We have
(
=
=
1
3
2 3
4 5
4 5
2 7
6
8
7
9
8
6
9
1
(13579)(24)(68).
Here we have one cycle of length 5 and two cycles of length 2. Hence
|| = l. c. m.(5, 2, 2) = 10.
Scal(n, R) =
0
0
: R, = 0
1 0 0
0 2 0
D(n, R) =
0
n
19
: i R, 1 n = 0
2
Tu (n, R) =
: i R, 1 n = 0
0
n
(b) The subgroup of upper unitriangular
U Tu (n, R) =
matrices:
1
Similarly, we have the subgroups of lower triangular and lower unitriangular matrices Tl (n, R) and U Tl (n, R).
Note:
U Tu (n, R)
SL(n, R)
U Tu (n, R)
Tu (n, R)
D(n, R)
Tu (n, R)
Scal(n, R) and D(n, R) are abelian for all n 1. Tu (n, R) is not abelian for all
n 2, and U Tu (n, R) is not abelian for all n 3.
All these subgroups may also be dened for the general linear groups over Q,C
and Zp . Of course, over Zp these subgroups will be nite, and it is a nice exercise
to work out their orders. For example,
|D(n, Zp )| = (p 1)n ,
|U Tu (n, Zp )| = p
n(n1)
2
CENTRALIZERS
Let G be an arbitrary group, and a G. Consider the set
C(a) = {g G : ga = ag},
i.e. the set of all elements that commute with the given (xed) element a.
Example: Let G = S4 , a = (1234). Then (13)(24) C(a), since
(1234)(13)(24) = (13)(24)(1234) = (1432),
but (123)
/ C(a) since (1234)(123) = (123)(1234). Indeed, we have (1234)(123) =
(1324), but (123)(1234) = (1342).
RALPH STOHR
20
and
(123) C((123)).
Also, (123)(132) = (132)(123) = e, so (132) C((123)). But the three transpositions do not commute with (123): (123)(12) = (13) but (12)(123) = (23),
(123)(13) = (23) but (13)(123) = (12) and (123)(23) = (12) but (23)(123) = (13).
Hence
C((123)) = {e, (123), (132)} = (123).
and
m1 (1) = m.
21
Now
(m) =
( m1 (1))
= ( m1 )(1)
= ( m1 )(1)
(since C( ))
= m1 ((1))
= m1 (k)
= m1 ( k1 (1))
= ( m1 k1 )(1)
= ( k1 m1 )(1)
= k1 ( m1 (1))
= k1 (m).
as required. Hence , so C( ) and hence C( ) = .
(
)
1 1
(c) Let A =
GL(2, R). What is C(A)?
1 0
(
)
a b
A 2x2 matrix B =
is in C(A) if and only if AB = BA and B is
c d
invertible.
Now: AB = BA is the same as
(
)(
) (
)(
)
1 1
a b
a b
1 1
=
1 0
c d
c d
1 0
Here
(
LHS =
Again
(
LHS =
a+c
a
b+d
b
a+c
a
b+d
b
(
RHS =
(
RHS =
Now, LHS=RHS i
a+c = a+b
a
= c+d
b+d
= a
= c.
This holds i
b=c
So AB = BA i
and
(
B=
a = b + d.
b+d b
b
d
a+b
c+d
a
c
a+b
c+d
a
c
)
.
)
.
RALPH STOHR
22
Hence
((
C
1 1
1 0
(
(d) Let D =
))
{(
=
b+d b
b
d
}
: b, d R, (b + d)d b = 0 .
2
)
1 0
GL(2, R). What is C(D)? Again, equating
0 2
(
)(
) (
)(
)
1 0
a b
a b
1 0
=
0 2
c d
c d
0 2
we get
(
LHS =
a
2c
b
2d
(
RHS =
a
c
2b
2d
)
.
THE CENTRE
Let G be an arbitrary group. The centre of G is the set
Z(G) = {g G : gx = xg x G}.
In other words, the centre of G is the set all those elements that commute with all
other elements in G.
Fact: The centre Z(G) is a subgroup of G.
Proof. Clearly, e Z(G), so Z(G) = , and we can apply the Subgroup
Criterion. Now, if g, h Z(G), i.e. gx = xg and hx = xh for all x G, then
gh Z(G) since for all x G we have
(gh)x = g(hx) = g(xh) = (gx)h = (xg)h = x(gh),
i.e. gh Z(G).
Also, if g Z(G), then g 1 Z(G) since for all x G we have
g 1 x = g 1 xe = g 1 xgg 1 = g 1 gxg 1 = exg 1 = xg 1 .
Hence Z(G) satises the two conditions of the Subgroup Criterion, so Z(G)
G.
Some obvious facts about the centre:
In any abelian group G, Z(G) = G.
In fact, a group G is abelian if and only if Z(G) = G.
For all a G, Z(G) C(a).
In fact,
Z(G) =
C(a).
aG
23
Examples: Again, since groups of numbers and the groups Zn are abelian, they
just coincide with their centres. However, things are dierent for symmetric groups
and matrix groups.
(a) Consider S3 = {e, (12), (13), (23), (123), (132)}. Here we have
(12)(13) =
(13)(12)
(12), (13)
/ Z(S3 )
(23)(123) =
(123)(23)
(23), (123)
/ Z(S3 )
(23)(132) =
(132)(23)
(132)
/ Z(S3 ).
Hence,
Z(S3 ) = {e}.
n 3.
(
RHS =
a+c b+d
c
d
c = 0 & a = d.
RALPH STOHR
24
)
a b
. But to be a
So if a matrix is in the centre, it must be of the form
0 a
central matrix, it must commute with all matrices in Gl(2, R). In particular, we
must have
)(
) (
)(
)
(
1 0
1 0
a b
a b
=
.
0 a
1 1
1 1
0 a
(
Here
LHS =
a+b b
a
a
(
RHS =
a
b
a a+b
)
.
b = 0.
a 0
, i.e. it is a scalar matrix. Of
0 a
course we must have a = 0 for the matrix to be invertible. So we have proved that
Z(Gl(2, R)) = Scal(2, R).
Alternatively, we could have used our earlier examples of centralizers to show
that Z(Gl(2, R)) = Scal(2, R). Recall that
((
)) {(
)
}
1 0
a 0
C
=
: a, d R, ad = 0 = D(2, R).
0 2
0 d
Hence a central matrix is of the form
and that
C
((
1 1
1 0
))
{(
=
b+d b
b
d
}
: b, d R, (b + d)d b = 0 .
2
Any matrix in the centre must belong to both of these centralizers, i.e. it must
be in the intersection of the two centralizers, which is the group of scalar matrices
Scal(2, R).
Our rst proof, however, can nicely be modied to establish a more general
result, namely that Z(Gl(n, R)) = Scal(n, R).
25
4. Cyclic Groups
Definition. A group G is called cyclic, if there exists an element a G such
that G = a.
In other words, G is cyclic if all its elements are powers of a single element. If
G = a, then a is called a generator for G.
Examples: (a) Z is an innite cyclic group:
Z = 1 = 1
26
RALPH STOHR
27
Examples:
(a)
Z 3 = {. . . , 6, 3, 0, 3, 6, 9, . . .}.
{. . . , 6, 3, 0, 3, 6, 9, . . .}, if a 0 mod 3;
a + 3 =
{. . . , 5, 2, 1, 4, 7, 10, . . .}, if a 1 mod 3;
(2)
{0, 5, 10},
{1, 6, 11},
k + 5 =
{4, 9, 14},
(3)
if k 0 mod 5;
if k 1 mod 5;
if k 4 mod 5.
Here we get
e(12)
(12)(12)
(13)(12)
(23)(12)
(123)(12)
(132)(12)
(23)(12) =
RALPH STOHR
28
xH = yH
xH yH = .
or
x = yh2 h1
1 .
and hence
(since h2 h1
1 h3 H.).
[Z15 : 5] = 5,
[S3 : (12)] = 3.
The examples also illustrate what we know from the Corollary, namely, that any
group G is the disjoint union of the left cosets of H in G.
Lemma 5.2. Let H be a nite subgroup of a group G. Then |gH| = |H| for all
g G.
Proof. Let
H = {h1 , h2 , . . . , hn }
and assume that the hi are distinct as listed. Then
gH = {gh1 , gh2 , . . . , ghn }.
29
But these elements too are distinct as listed. Indeed, if ghi = ghj , then hi = hj by
the Cancelation Lemma. Hence gH has exactly n elements, so |gH| = |H|.
LAGRANGES THEOREM. Let G be a nite group and H G. Then
|G| = [G : H]|H|
where [G : H] is the index of H in G.
Proof. Let [G : H] = r and let
g1 H, g2 H, . . . , gr H
be the distinct left cosets of H in G. By the Corollary to Lemma 5.1, G is the
disjoint union of those cosets:
G = g1 H g2 H gr H
Since the union is disjoint we have
|G| = |g1 H| + |g2 H| + + |gr H|.