|
[Next][Prev] [Right] [Left] [Up] [Index] [Root]
Since the beginning of class field theory one of the core problems
has been to find defining equations for the fields. Although
most of the original proofs are essentially constructive, they rely on
complicated and involved computations and thus were not suited for
hand computations.
The method used here to compute defining equations is essentially
the same as the one
used for number fields. The main differences are due to the problem
of p-extensions in characteristic p where Artin-Schreier-Witt
theory is used, and the fact that the divisor class group is infinite.
AbelianExtension(D, U) : DivFunElt, GrpAb -> FldFunAb
Given an effective divisor D and a subgroup U of the ray class group,
(see RayClassGroup),
ClD of D such that the quotient ClD/U is finite,
create the extension defined by this data. Note that, at this point,
no defining equations are computed.
For a relative extension K/k of global function fields, compute the
maximal abelian subfield K/A/k of K/k as an abelian extension of
k. In particular, this function compute the norm group of K/k
as a subgroup of a suitable ray class group.
For a global function field K and a place p of K, compute the (a)
Hilbert class field of K as an abelian extension of K. This field
is characterised by being the maximal abelian unramified extension of K
where p is totally split.
WithAut: BoolElt Default: false
SetVerbose("ClassField", n): Maximum: 3
Given an abelian extension of function fields as created by
AbelianExtension,
compute
defining equations for the corresponding (ray) class field.
More precisely: Let Cl/U = ∏Cl/Ui be a decomposition
of the norm group of A
such
that the Cl/Ui are cyclic of prime power order. Then for each
Ui the function will compute a defining equation, thus A is
represented by the compositum.
If WithAut is true, the second sequence returned contains a
generating automorphism for each of the fields returned as the first
return value. If WithAut is false, the first (and only) return value is
a function field in non-simple representation.
First we have to create a function field, a divisor and a ray class group:
> k<w> := GF(4);
> kt<t> := PolynomialRing(k);
> ktx<x> := PolynomialRing(kt);
> K := FunctionField(x^3-w*t*x^2+x+t);
> lp := Places(K, 2);
> D := 4*lp[2]+2*lp[6];
> R, mR := RayClassGroup(D);
Let us compute the maximal extension of exponent 5 such that
the infinite place is totally split. This means that we
- -
- have to create a subgroup of R containing the image of the infinite
place
- -
- compute the fifth power of R
- -
- combine the two groups
and use this as input to the class field computation:
> inf := InfinitePlaces(K);
> U1 := sub<R | [x@@ mR : x in inf]>;
> U2 := sub<R | [5*R.i : i in [1..Ngens(R)]]>;
> U3 := sub<R | U1, U2>;
> A := AbelianExtension(D, U3);
> A;
Abelian extension of type [ 5 ]
Defined modulo 4*(t^2 + t + w, $.1 + w^2*t + 1) +
2*(t^2 + w^2*t + w^2, $.1 + w*t + w)
over Algebraic function field defined over
Univariate rational function field over GF(2^2) by
x^3 + w*t*x^2 + x + t
> FunctionField(A);
Algebraic function field defined over K by
$.1^5 + (w*K.1^2 + (w*t + w^2)*K.1 + (w^2*t^2 + t
+ w^2))*$.1^3 + ((t^2 + 1)*K.1^2 + w*t*K.1 +
(w*t^4 + w))*$.1 + (t^3 + w*t^2 + w^2*t)*K.1^2
+ (w*t^4 + w^2*t^2 + w^2)*K.1 + t^4 + w^2*t^3
+ w^2*t + w
> E := $1;
> #InfinitePlaces(E);
10
Which shows that all the places in inf split completely.
Now, suppose we still want the infinite places to split, but now we are
looking for degree 4 extensions such that the quotient of genus by
number of rational places is maximal:
> q, mq := quo<R | U1>;
> l4 := [ x`subgroup : x in Subgroups(q : Quot := [4])];
> #l4;
768
> A := [AbelianExtension(D, x@@mq) : x in l4];
> s4 := [<Genus(a), NumberOfPlacesOfDegreeOne(a), a>
> : a in A];
> Maximum([x[2]/x[1] : x in s4]);
16/5 15
> E := FunctionField(s4[15][3]);
> l22 := [ x`subgroup : x in Subgroups(q : Quot := [2,2])];
> #l22;
43435
Since this is quite a lot, we won't investigate them here further.
[Next][Prev] [Right] [Left] [Up] [Index] [Root]
|