In the present section, some elementary results are collected about finite fields, and some broad outlines of their proofs is given. A less cursive approach of these questions can be found in [Lidl, 9] or in [10].
Let
be a finite field of cardinal
. Let
be the (unique) ring homomorphism from
into
. Since
is finite,
cannot be the
ideal and, since
is an integral domain,
is a prime ideal. The (positive)
generator of that ideal is the so-called characteristic of the given field,
and it is to be noticed that any
contains some prime field
as subfield.
Let
be given in
, and
denotes now, the
mapping
defined as
.
Then again
is a group homomorphism, and since that mapping cannot
be injective, the ideal
is a
for some
.
That
is the so-called (multiplicative) order of
, and will
be denoted as
.
As a consequence, any
is a root of the polynomial:
,
which obviously belongs to
. Hence each and every
element of
is algebraic over
, and thus
is a subfield of the algebraic closure of
. In short:
.
Let be given an
. Since
is principal,
the set
is an ideal generated by an unique monic polynomial. That polynomial is obviously
irreducible and will be referred as the ``minimal polynomial for
''.
We denote it by
and put
.
Any
is a cyclic group [3], i.e., it exists at least one primitive
element
such that
. A sketch of the proof
is as follows :
Collecting all these results, we get that any finite field is algebraically
generated over its prime field
by some (proper) element
,
and thus is isomorphic to some quotient field of
.
As a consequence,
is a vector space over
, proving that
is a power of a prime, i.e.,
. Moreover
cyclic, and is thus multiplicatively generated by some (primitive) element
.
As a summary, we get:
The elementary binomial formula
comes to a nice result when
is taken equal to the characteristic of
. Since any prime
as the property to divide any of the
binomials
except for the ending ones, the formula shortens
to
. Thus the mapping
is a morphism of
into itself. But
implies
and
is injective
and therefore bijective.
The Frobenius mapping
induces an automorphism over
.
Let
be some element of
,
as minimal polynomial,
and
. Since the coefficients of
are in
,
they are
-invariant, and we get
.
Any
is therefore a root of
, and
since
, there are at most
different
.
On the other hand,
leads to
where
and
is a root of
. But
is
, implying
and
.
Therefore,
divides
: there are exactly
different
.
The elements of a
-sized
-orbit are the roots of an irreducible
polynomial of degree
over
.
Therefore
is the subfield of
that is fixed by
, leading to the following embedding result
.
That result can also be seen as resulting from the divisibility of one characteristic
polynomial by the other.
Let us now consider all the multiplicative subgroups of a given
.
Since that group is cyclic, it contains exactly one subgroup of cardinal
for each
that divides
. Some of them are a field-group
where
, but most of them are not. Therefore, in Fig. 1,
the subfields lattice (boldfaced) is a sublattice of the multiplicative groups
lattice.
The proper elements of such a
-sized subgroup are exactly the
roots of
, the cyclotomic polynomial relative
to the
exponent. As an example, when
, we get
and therefore the characteristic polynomial of
splits into
.
Moreover, since
describes the primitive elements of
,
it splits into two polynomials of degree 4, and since
describe proper elements of
, they split over
into
irreducible polynomials of degree 8.
The so-called trace of an element in
is defined as
.
Thus, for a proper element
of
,
is the sum of all the roots of the minimal polynomial of
. For an
of smaller degree, that sum is to be multiplied by the natural integer
and then reduced modulo
. Elementarily,
.
Given a primitive element
, there is a one to one correspondence
between the
and the
.
With the convention
, we get a bijection
from
towards
(since
can not be ordered as a group, the traditional
notation, ``
'', is rather confusing).
That bijection
can be used to carry the field structure of
onto
. Obviously, the image of
inside
is
inside
. But, since integer arithmetic modulo
is far easier than computations of polynomials modulo
over
, it is of great interest to consider the isomorphic image in
of the first law (addition) in
.
Let us denote by
that first law over
. From
,
we get
,
and therefore the whole
table is known when the function
is known. The so-called Zech logarithm is nothing but precisely that function
.