Matrix Groups

Among the most important examples of groups are groups of matrices. The textbook
briefly discusses groups of matrices in Chapter 2, and then largely forgets about them.
These notes remedy this omission.

Matrix Groups over Fields

As we have seen, it is possible to consider matrices whose entries are elements of any
field F . Common choices for F include the field R of real numbers, the field C of
complex numbers, and finite fields such as Zp for p prime.

Definition: Matrix Group

A matrix group over a field F is a set of invertible matrices with entries in F that
forms a group under matrix multiplication.

Note that the matrices in a matrix group must be square (to be invertible), and
must all have the same size. Thus there are 2 2 matrix groups, 3 3 matrix groups,
4 4 matrix groups, and so forth. The size of the matrices is sometimes referred to
as the degree of the matrix group.
Because the group operation for a matrix group is matrix multiplication, the
identity element of a matrix group is always the n n identity matrix, and inverses
in a matrix group are just the usual inverse matrices.
The most important matrix groups are the general linear groups.

Definition: General Linear Group

Let F be a field, and let n N. The general linear group of degree n over F ,
denoted GL(n, F ), is the group of all invertible n n matrices with entries in F .
 Matrix Groups 2

Since a matrix is invertible if and only if its determinant is nonzero, GL(n, F ) can
also be defined as the group of all n n matrices with entries in F having nonzero
determinant. (Note that the determinant of a matrix with entries in F is by definition
an element of F .)

EXAMPLE 1 GL(2, Z2 )
If F is an infinite field such as R or C, then the general linear group GL(n, F ) has
infinite order. However, if F is a finite field, then GL(n, F ) is a finite group.
For example, consider all 2 2 matrices over the field Z2 . There are sixteen such
matrices, since each of the four entries can be either 0 or 1. Of these sixteen matrices,
exactly six of them have nonzero determinant:
           
1 0 0 1 1 0 0 1 1 1 1 1
, , , , , .
0 1 1 0 1 1 1 1 0 1 1 0

Thus GL(2, Z2 ) is a six-element group. It is easy to check that GL(2, Z2 ) S3 . 

The general linear group GL(n, F ) is the most general matrix group, in the same
way that Sn is the most general permutation group. In particular, every matrix group
is just a subgroup of some GL(n, F ).
The most important subgroup of GL(n, F ) is the special linear group.

Definition: The Special Linear Group

Let F be a field, and let n N. The special linear group of degree n over F is
defined as follows

SL(n, F ) = {A GL(n, F ) | det(A) = 1}.

Note that SL(n, F ) is indeed a subgroup of GL(n, F ). In particular, it is always

nonempty (since the identity matrix has determinant 1), and it is closed under mul-
tiplication and inverses.
The following examples discuss a few other important matrix groups.

EXAMPLE 2 Diagonal Matrices

A diagonal matrix is a square matrix whose nonzero entries are all along the main
diagonal:
1 0 0
0 2 0

.. .. . . ..
. . . .
0 0 n
 Matrix Groups 3

It is not hard to see that the set of all diagonal matrices in GL(n, F ) forms a subgroup.
We shall verify this using the Two-Step Subgroup Test.

1. First, observe that the product of two diagonal matrices is again diagonal, with

1 0 0 1 0 0 1 1 0 0
0 2 0 0 2 0 0 2 2 0
=

.. .. . . .. .. .. . . .. .. .. .. ..
. . . . . . . . . . . .
0 0 n 0 0 n 0 0 n n

Thus the set of diagonal matrices is closed under multiplication.

2. Similarly, the inverse of a diagonal matrix is obtained by taking the multiplica-

tive inverse of each diagonal entry:
1 1
1 0 0 1 0 0
0 2 0 0 1 0
2
.. = .. .

.. .. . . .. .. ..
. . . . . . . .
0 0 n 0 0 1
n

Thus the set of diagonal matrices is closed under inverses. 

EXAMPLE 3 Triangular Matrices

An upper triangular matrix is an n n matrix that has zeroes below the main
diagonal:
a11 a12 a1n

0
a22 a2n

. .. .. ..
.. . . .
0 0 ann
Such a matrix is invertible if and only if all of its diagonal entries are nonzero.
Though it is slightly difficult to prove, the inverse of an upper triangular matrix
is upper triangular, and the product of two upper triangular matrices is again upper
triangular. Thus the set of all upper triangular matrices in GL(n, F ) form a matrix
group.
The transpose of an upper triangular matrix is called a lower triangular matrix.
Everything that works for upper triangular matrices works just as well for lower
triangular matrices. In particular, the set of all lower triangular matrices in GL(n, F )
also form a matrix group. 
 Matrix Groups 4

Linear Transformations
For the remainder of these notes, we will only be considering vectors and matrices
over the real numbers.
Let Rn denote n-dimensional Euclidean space, i.e. the set of all column vec-
tors with n components. For example, R2 is the Euclidean plane, and R3 is three-
dimensional Euclidean space.
If A is an n n matrix and v Rn , the product Av is again an element of Rn .
This lets us think of any n n matrix as a function from Rn to itself.

Definition: Linear Transformation

Let A be an n n matrix. The linear transformation associated with A is the
function T : Rn Rn defined by

T (v) = Av

for all v Rn .

EXAMPLE 4 Linear Transformations of R2

Consider the following three matrices:
     
1 0 2 0 0 1
A1 = , A2 = , A3 = .
0 1 0 2 1 0
These matrices define three different linear transformations of R2 .
The transformation T1 : R2 R2 corresponding to A1 is given by the formula
      
x 1 0 x x
T1 = = .
y 0 1 y y
This transformation negates the y-coordinate of each point, and therefore reflects the
plane R2 across the x-axis.
The transformation T2 : R2 R2 corresponding to A2 is given by the formula
      
x 2 0 x 2x
T2 = = .
y 0 2 y 2y
Geometrically, this transformation scales each vector by a factor of 2. Therefore, T2
corresponds to a dilation of the plane by a factor of 2 centered at the origin.
The transformation T3 : R2 R2 corresponding to A3 is given by the formula
      
x 0 1 x y
T3 = = .
y 1 0 y x
It is easy to check that this is a 90 counterclockwise rotation of the plane around
the origin. 
 Matrix Groups 5

You may be surprised that the reflection, dilation, and rotation in the last ex-
ample were linear transformations. This brings up several questions. First, is every
reflection, dilation, or rotation of the plane a linear transformation? And what about
other sorts of transformations, such as translations?
The following theorem places some restriction on the possible linear transforma-
tions.

Theorem 1 Zero is Fixed

If T : Rn Rn is a linear transformation, then T (0) = 0, where 0 denotes the
zero vector in Rn .

PROOF Let T : Rn Rn be a linear transformation, and let A be the corresponding

n n matrix. Then T (v) = Av for all v Rn , so T (0) = A0 = 0. 

Geometrically, this theorem says that any linear transformation must fix the origin
in Rn . Thus a translation cannot a linear transformation, and a reflection, dilation,
or rotation can only be a linear transformation if it fixes the origin.
Our next theorem gives a complete geometric classification of linear transforma-
tions. This is a surprisingly difficult theorem to prove, and we shall not attempt to
do so here.

Theorem 2 Geometric Characterization

Let n 2, and let T : Rn Rn be any function. Then T is a linear transforma-
tion if and only if it satisfies the following conditions:

1. T fixes the origin, i.e. T (0) = 0.

2. For every straight line L Rn , the image of L under T is also a straight

line.

Using this theorem, we can immediately see that many familiar geometric trans-
formations are in fact linear transformations. These include:

Any reflection of R2 across a line through the origin.

Any reflection of R3 across a plane through the origin, or more generally any
reflection of Rn across an (n 1)-dimensional subspace.
 Matrix Groups 6

Any rotation of R2 about the origin.

Any rotation of R3 that fixes the origin.

Any dilation of Rn centered at the origin.

Transformation Groups
As we have seen, some of the most important geometric transformations of Rn are
linear transformations. As a result, the symmetry group of a geometric object can
sometimes be thought of as a group of linear transformations.

Definition: Transformation Group

A transformation group is a group whose elements are linear transformations,
and whose operation is composition.

We have already seen a few examples of transformations groups:

The dihedral groups are transformation groups. In particular let P be a regular

n-gon in the plane centered at the origin. Then each rotation or reflection of P
is a linear transformation, so the resulting dihedral group Dn is a transformation
group.

Similarly, let P be any polyhedron in R3 (such as a tetrahedron or cube) centered

at the origin. Then every symmetry of P is a linear transformation, so the
symmetry group of P is a transformation group.

The following theorem can be very helpful for working with transformation groups.

Theorem 3 Compositions and Inverses of Transformations

Let A and B be n n matrices, and let TA and TB be the corresponding linear

transformations. Then:

1. The composition TA TB is a linear transformation, corresponding to the

matrix AB.

2. If A is invertible, then TA is bijective, and (TA )1 is the linear transformation

corresponding to the matrix A1 .
 Matrix Groups 7

PROOF Recall that TA (v) = Av and TB (v) = Bv for any vector v Rn . Then
(TA TB )(v) = TA (TB (v)) = TA (Bv) = A(Bv) = (AB)v
for all v Rn , which proves that TA TB is the linear transformation corresponding
to AB.
For part (2), let TA1 be the linear transformation corresponding to A1 , and let
TI be the linear transformation corresponding to the n n identity matrix. Since
AA1 = A1 A = I, it follows from part (1) that TA TA1 = TA1 TA = TI . But TI
is the identity function on Rn , and therefore TA1 an the inverse function for TA . 

This theorem lets us work with transformation groups as though they were groups
of matrices. Indeed, every transformation group is isomorphic to a group of matrices.

Corollary 4 Matrix Representation of Transformation Groups

Let G be a transformation group, and let H be the corresponding set of n n

matrices. Then H is a subgroup of GL(n, R), and G and H are isomorphic.

PROOF Left to the reader. 

EXAMPLE 5 A Dihedral Group

Let D4 be the group of symmetries of the square [1, 1] [1, 1] in R2 . Recall that
D4 = {e, r, r2 , r3 , s, rs, r2 s, r3 s},
where r is a 90 counterclockwise rotation about the origin, and s is a reflection
across the x-axis. Note that each of the elements of D4 is a linear transformation,
and therefore D4 is a transformation group.
As seen Example 4, the transformations r and s correspond to the following pair
of matrices:    
0 1 1 0
R= and S= .
1 0 0 1
By multiplying R and S, we can find matrices corresponding to each of the elements
of D4 :
       
1 0 0 1 2 1 0 3 0 1
I= R= R = R =
0 1 1 0 0 1 1 0
       
1 0 0 1 2 1 0 3 0 1
S= RS = R S= R S=
0 1 1 0 0 1 1 0
 Matrix Groups 8

In particular, each of the matrices R, R2 , R3 corresponds to a rotation of the plane,

while each of the matrices S, RS, R2 S, R3 S corresponds to a reflection of the plane
across a certain line through the origin. Together, these eight matrices form a sub-
group of GL(2, R) that is isomorphic to D4 . 

Finding the Matrix for a Transformation

For our matrix representation approach to be useful, we need a method to find the
matrix that corresponds to a given linear transformation. We shall use a certain trick
which is often useful for finding the columns of an unknown matrix.
Recall that the standard basis vectors for Rn , denoted e1 , e2 , . . . , en are the
columns of the n n identity matrix:

1 0 0
0 1 0
e1 = .. , e2 = .. , . . . , en = .. .

. . .
0 0 1
The trick we will use is this: if A is any matrix and ei is a standard basis vector, then
the product Aei is just the ith column of A. For example,

1 2 3 0 3
4 5 6 0 = 6.
7 8 9 1 9

Theorem 5 Formula for the Matrix of a Linear Transformation

Let T : Rn Rn be a linear transformation. Then the columns of the matrix
corresponding to T are the vectors

T (e1 ), T (e2 ), ..., T (en ),

where e1 , e2 , . . . , en denote the standard basis vectors for Rn .

PROOF Let A be the n n matrix corresponding to T , so that T (v) = Av for all

v Rn . Then T (ei ) = Aei for each standard basis vector ei . But Aei is just the ith
column of A, and therefore the ith column of A is T (ei ). 

EXAMPLE 6 The Matrix for a Reflection

Let T : R2 R2 be reflection across the line y = x. It is clear from the geometry
 Matrix Groups 9

that        
1 0 0 1
T = and T = ,
0 1 1 0
 
0 1
so the matrix for T is . 
1 0

EXAMPLE 7 The Matrix for a Rotation

Let T : R2 R2 be a counterclockwise rotation about the origin by an angle of . A
little basic trigonometry shows that
cos( + 90 )
         
1 cos 0 sin
T = and T = =
0 sin 1 sin( + 90 ) cos
 
cos sin
Therefore, the matrix for T is . 
sin cos

The method of Theorem 5 also works for linear transformations in three dimen-
sions.

EXAMPLE 8 A Rotation in Three Dimensions

Let L be the line y = x in the xy-plane, and let T be the 180 rotation of R3 about
the line L. It is not hard to see that

1 0 0 1 0 0
T 0 = 1 ,
T 1 = 0 ,
and T 0 = 0 ,
0 0 0 0 1 1

0 1 0
so the matrix for T is 1 0 0. 
0 0 1

Using the method of Theorem 5, we can find matrices corresponding to the el-
ements of any transformation group. In most cases, the fastest approach is to use
Theorem 5 to find matrices corresponding to the elements of a generating set, and
then multiply to obtain the remaining matrices.

EXAMPLE 9 Dihedral Group of a Triangle

Let D3 be the symmetry group of an equilateral triangle centered at the origin of R2 ,
with one of its vertices along the x-axis. Then

D3 = {e, r, r2 , s, rs, r2 s},

 Matrix Groups 10

where r is a 120 counterclockwise rotation about the origin, and s is a reflection

across the x-axis. According to the formula in Example 7, the matrix corresponding
to r is 
cos 120 sin 120
  
1 1 3
R = = .
sin 120 cos 120 2 3 1
 
1 0
Combining this with the matrix S = , we obtain the following six matrices:
0 1
     
1 0 1 1 3 1 1 3
I = R = R = 2
0 1 2 3 1 2 3 1
    
1 0 1 1 3 1 1 3
S= RS = 2
R S =
0 1 2 3 1 2 3 1

This six-element matrix group is isomorphic to the dihedral group D3 . 

EXAMPLE 10 Rotations of a Cube

Let G be the group of rotations of the cube [1, 1] [1, 1] [1, 1] in R3 . Note
that, among other things, G includes 90 rotations about the x and y axes. The
corresponding 3 3 matrices are

1 0 0 0 0 1
Rx = 0 0 1 and Ry = 0 1 0 .
0 1 0 1 0 0

Using matrix multiplication, we can combine Rx and Ry to obtain the group of 24

possible rotations of the cube:
           
1 0 0 0 1 0 1 0 0 0 0 1 0 0 1 0 1 0
0 1 0 , 1 0 0 , 0 0 1 , 0 1 0 , 1 0 0 , 0 0 1
0 0 1 0 0 1 0 1 0 1 0 0 0 1 0 1 0 0
           
1 0 0 0 1 0 1 0 0 0 0 1 0 0 1 0 1 0
0 1 0 , 1 0 0 , 0 0 1 , 0 1 0 , 1 0 0 , 0 0 1
0 0 1 0 0 1 0 1 0 1 0 0 0 1 0 1 0 0
           
1 0 0 0 1 0 1 0 0 0 0 1 0 0 1 0 1 0
0 1 0 , 1 0 0 , 0 0 1 , 0 1 0 , 1 0 0 , 0 0 1
0 0 1 0 0 1 0 1 0 1 0 0 0 1 0 1 0 0
           
1 0 0 0 1 0 1 0 0 0 0 1 0 0 1 0 1 0
0 1 0 , 1 0 0 , 0 0 1 , 0 1 0 , 1 0 0 , 0 0 1 
0 0 1 0 0 1 0 1 0 1 0 0 0 1 0 1 0 0