*: GxG -->
G ("multiplication"); a neutral element ("unit")
e and an inverse i(x)=x-1 for
each element x in G. The rules are:
x*(y*z)=(x*y)*z;i(x)*x = x*i(x) = e;x*e = e*x = x.The group is called abelian if the operation is commutative:
x*y = y*x
Subgroup is a subset containing e and
closed under * and i.
Examples:
(Z, +, 0);(Q\{0}, *, 1);Sn - the group of permutations of
n elements; operation: composition; unit: identity
map.A ring is (R,+,0,*,1) such that (R,+,0)
is an abelian group and (R\{0},1,*) is a commutative
"semigroup", i.e., a "group without inverse". An additional condition
is distributivity:
x*(y+z) = x*y + x*z.
Subring is a subset containing 1 and 0 and closed under
+,-,*.
Examples:
(Z,+,*);R[x];n - Z/nZ:
{0,1,...,n-1} with operations x+y := x+y mod
n; x*y := x*y mod n etc.Exercise: Prove that Z/nZ is a ring.
A field (K,+,0,*,1) is a ring such that
(K\{0},*,1) is a group (called multiplicative group of a
field and denoted K*); i.e., every nonzero
element has a multiplicative inverse. Subfield is a subset containig
0 and 1 and closed under +,-,*,/.
Examples:
(Q,+,0,*,1);(R,+,0,*,1);R(x);p: Z/pZ =
Fp.Exercise: Prove that Z/pZ is indeed a
field. Why Z/nZ for a non-prime n is not a
field?
Problem:
Q is the idenity map.R is the idenity map.#.x of a group G: the
smallest integer number n such that
xn=1. It is also equal to the order of the
subgroup generated by the element. The order of an element divides
the order of the group (if both are finite), since the following is
true for any subgroup H: #G = #H * #(G/H)
(Prove it).Q (rationals) and
Fp = Z/pZ (for prime p).Q and p if its simple subfield is
Fp. Denoted: char.
3. Prove that if char K = p then for any
- the sum x - element of K( x +
... + x ) (p items) is 0.
p: 1+1+...+1=0
(p 1's). Prove that p is prime.K. It contains 0 and 1.
Consider the sequence an,
s.t. a0=0; a1=1;
a2=1+1; and, in general,
an+1=an+1. (Sloppily we could
write it as an=n*1; remember, the integer
number n is not an element of K). Either
all elements in this sequence are different, and then the subring
{an; -an} is isomorphic to
Z (then K contains a subfiels isomorphic
to Q and its characteristic is 0), or there exists a
smallest k such that ap=0.
Then {al | l=1..p} is isomorphic to
Z/pZ. If p=nm is not a prime, we have
an*am=0 and K is
not a field. Therefore any field contains a subfield isomorphic to
Q or Fp, i.e., up to
isomorphism, there are the only prime fields.p units is 0 - by the definition of the
characteristic: 1 lies in the simple subfield
Fp.Fp is infinite, but its characteristic is
finite and equal to p.
Fq;
#Fq=q. Let char
Fq=p. Prove that q=pk for
some integer k.
Fq as a vector space over its
subfield Fp (i.e. Fp
is the field of scalars). Let its dimension be k. (Why
is the dimension finite?) What is known about the structure of a
finite dimensional vector space? (Coordinates, basis...)
Fp (its simple
subfield), Fq (a finite
q-element field) is finite-dimensional (since it's
finite), and thus isomorphic as a vector space to
(Fp)k, with k=dim
Fq. Thus q = #Fq =
pk.Consider a finite field Fq; char Fq=p;
q=pk; and let F: Fq -->
Fq be the following transformation:
F(x)=xp.
F is a "field automorphism", i.e.,
F(x-y)=F(x)-F(y); F(x/y)=F(x)/F(y).
p -
prime, k=1,...,p-1; p divides
C(p,k) (the binomial coefficient).
The statement F(x/y)=F(x)/F(y) follows from the
usual properties of powers. Next,
(x-y)p=xp + (-y)p +
Sum1p-1 C(p,k)xk
yp-k.
p divides C(p,k) for
k=1..p-1 and prime p, since we know that
C(p,k)=p!/k!(p-k)!, and p divides the
numerator but does not divide the denominator. Thus the "Sum" term
is 0 and F(x-y)=xp + (-y)p. If
p=2 then A+A=0 and A=-A thus
F(x-y)=x2 + y2 = x2 -
y2. If p is odd then
(-y)p = -yp.
F(x+y)=F(x)+F(y);
F(xy)=F(x)F(y); F(-x)=-F(x); F(x-1)=F(x)-1;
F(1)=1; F(0)=0. Prove it! You might think about which set of
conditions other than the one from the problem above will gurantee
that the map is an automorophism.
F(x)=x iff (if and only if) x
lies in the simple subfield Fp.
Xp-X and its roots.
X in Fp,
xp=x (see below). Since
Xp-X cannot have more than p
roots in Fq (and all p elements
of Fp are its roots), no element of
Fq\Fp can be a root.x of field
Fq we have xq=x;
i.e., Pq(x)=0.tq-1=1 and for any l<q-1;
tl != 1. This means that the multiplicative group
Fq* of the field
Fq is isomorphic to Z/(q-1)Z
(is "cyclic": each nonzero element of the field is some power of
the element t, which is called "generator").Fq* = Fq \ {0}
with the operation of multiplication. Take an arbitrary element and
multiply it by itself several times. You cannot always get a
different element, can you? This treats the first part, since the
order of an element must divide the order of the group
q-1.o(x) to be the order of x, i.e., the
smallest number l such that
xl=1. Let o = max {o(x) | x in
Fq}. Prove that for any x we have
that o(x) divides o, i.e.,
xo=1. (This is true since
o(xy)<=lcm(o(x),o(y)); lcm = Least Common Multiple).
Then if o<q-1, one must have a polynomial
P(x)=xo-1 with more than o
roots (namely, q-1).
Is it possible in a field? Why?Consider the polynomial
P(x)=Pq(x)/Pp(x) (why is it a
polynomial?)
Fp.
P(x) is the product of all irreducible polynomials of
degree k over Fp. (recall that
q=pk).
z in Fq\Fp.
Let z0=z; and
zl+1=F(zl) (the Frobenius
automorphism), i.e.,
zl=zpl; thus
zk=z=z1. Therefore the polynomial
Pz(x)=
(x-z0)(x-z1)...(x-zk-1) is
invariant under the Frobenius automorphism, thus its coefficients lie
in Fp. These polynomials are irreducible
over Fp. (Why? Apply the Frobenius
automorphism to its alleged irreducible factor. It cannot change
because of its coefficients - they lie in Fp
and thus are invariant under the Frobenius automorphism - and it has
to change because of its roots - they are not the whole set {z,
F(z)=zp, F(F(z))=zp2, ...,
F(F(...(F(z))...)) = zpk-1} and thus
must change under the Frobenius automorphism). Any other irreducible
polynomial of degree k has to be one of these. (It
cannot have roots in Fp - then it's reducible
- so it has roots in the algebraic closure of
Fp, but with a root z it has to
have F(z), F(F(z)), F(F(F(z))) etc. as roots, and it
cannot have more than k roots, so
zpk=zq=z - and thus
z lies in Fq.)f: Fq -->
Fq there exist the unique polynomial
Pf of degree no more than q-1
such that for any x in Fq we
have Pf(x)=f(x). How many such functions are
there?
Aut(Fq):Fq. The operation is the composition, the
unit is the identity map, the inverse is the inverse map.
Aut(Fq).
Aut(Fq) is isomorphic to
Z/kZ, where q=pk; and the
generator of Aut(Fq) is the Frobenius
automorphism F.
F as an element of
Aut(Fq) is k. Then prove that
there are only k elements in this group.
Suppose the order of F is l<k. Then
Fl = id and for each element x of
the field Fq we have
xpl = x, i.e., any element of the
field Fq is a root of the polynomial
xpl-x of degree
pl<pk=q which is
impossible. Therefore for i=1,...,k-1 we have
Fi != id and
F0=Fk=id (since
Fk(x)=xq=x). This also implies that
the set {Fl | l=0,1,...,k-1} is a subgroup of
Aut(Fq) isomorphic to Z/kZ.
Next, consider the generator t of the field
Fq. Recall that for each nonzero element
x of the field Fq there is a
unique number 0<=l<=q-1 such that
x=tl. Suppose we have an automorphism
w of the field Fq and its value
on the generator t is w(t)=ta for
some 0<=a<=q-1. Then w(x) = w(tl)
= w(t)l = {ta}l = tal =
{tl}a = xa. Thus the
automorphism w is determined by the image of the
generator and furthemore, it can be represented in the form x
--> xa. Now the only thing left to be proven is
that a must be a power of p.
Here we again use the Newton's binomial. Since w is
an automorphism, we have the following identity for all x
from the field Fq: w(x+1) = w(x) + w(1)
= w(x) + 1. Therefore we have (x + 1)a =
xa + 1 or that the polynomial (x +
1)a - xa - 1 is identical to 0. Since its
degree is less than q, the coefficients must be 0, and
the coefficients are the binomial coefficients C(a,l) for
l=1,...,a-1. They are 0 in the field
Fq if and only if p divides each
of them. Suppose that
a=pm+b<pm+1. Careful
consideration of C(a,b) shows that it is not divisible by
p unless b=0. The reader is encouraged to
actually perform the "careful consideration".
| Sam Steingold<sds@gnu.org> | created: 1995-01-01 |