Automorphism Groups

Automorphisms of an algebraic field and the group they form can be calculated. Furthermore, field invariants that relate to the automorphism group can be determined.

Automorphisms(F): FldAlg -> [ Map ]
Abelian          : BoolElt                    Default: false
AutomorphismGroup: Verbose                    Default: 3

Given an algebraic field \(F\), return the automorphisms of \(F\) as a sequence of maps. If the extension is known to be abelian, the parameter Abelian should be set to true in which case a much more efficient algorithm [AK99, Klu97] will be employed. If \(F\) is not a normal extension, the automorphisms are obtained by a variation of the polynomial factorisation algorithm.

AutomorphismGroup(F): FldAlg -> GrpPerm, PowMap, Map
Abelian          : BoolElt                    Default: false
UseGalois        : BoolElt                    Default: false
AutomorphismGroup: Verbose                    Default: 3

Given an algebraic field \(F\), that is either a simple normal extension of \({\mathbb{Q}}\) or simple abelian extension of \({\mathbb{Q}}\), return the automorphism group \(G\) of \(K\) as a permutation group of degree \(n\), where \(n\) is the degree of the extension. If the extension is known to be abelian, the parameter Abelian should be set to true in which case a much more efficient algorithm [AK99, Klu97] will be employed. If UseGalois is set, GaloisGroup will be called and the result will be derived from that. If \(F\) is not a normal extension of \({\mathbb{Q}}\) an error will occur. In addition to returning \(G\), the function also returns the power structure \(Aut\) of all automorphisms of \(F\), and the transfer map \(\phi\) from \(G\) into \(Aut\).

Example: Automorphisms of a Quartic Field (ex-7b9102)

Run in calculator (opens in a new tab)

We consider the extension obtained by adjoining a root of the irreducible polynomial \(x^4 - 4x^2 + 1\) to \({\mathbb{Q}}\).

> Q := RationalField();
> R<x> := PolynomialRing(Q);
> K<w> := NumberField(x^4 - 4*x^2 + 1);
> A := Automorphisms(K);
> A;
[
    Mapping from: FldNum: K to FldNum: K,
    Mapping from: FldNum: K to FldNum: K,
    Mapping from: FldNum: K to FldNum: K,
    Mapping from: FldNum: K to FldNum: K
]
> for phi in A do phi(w); end for;
w
w^3 - 4*w
-w^3 + 4*w
-w

Taking the same field \(K\) we use instead the function AutomorphismGroup:

> G, Aut, tau := AutomorphismGroup(K);
> for x in G do tau(x)(w); end for;
w
w^3 - 4*w
-w^3 + 4*w
-w

Run in calculator (opens in a new tab)

AutomorphismGroup(K, F): FldAlg, FldAlg -> GrpPerm, PowMap, Map
AutomorphismGroup(K, F): FldRat, FldAlg -> GrpPerm, PowMap, Map
AutomorphismGroup(K, F): FldAlg, FldRat -> GrpPerm, PowMap, Map

Computes the group of \(K\) automorphisms of \(F\) as a permutation group together with a list of all automorphisms and a map between the permutation group and explicit automorphisms of the field.

This function computes the automorphism group of \(F\) over \({\mathbb{Q}}\) first.

DecompositionGroup(p): RngIntElt -> GrpPerm
DecompositionGroup(p): RngOrdIdl -> GrpPerm

For an ideal \(p\) of the maximal order of some absolute normal field \(F\) with group of automorphisms \(G\), compute the decomposition group, i.e. the subgroup \(U\) of the automorphism group such that:

\[U := \{ s \in G | s(p) = p \}\]

If \(F\) is not a normal extension of \({\mathbb{Q}}\) an error will occur.

RamificationGroup(p, i): RngOrdIdl, RngIntElt -> GrpPerm

For an ideal \(p\) of the maximal order \(M\) of some absolute normal field \(F\) with group of automorphisms \(G\), compute the \(i\)-th ramification group, i.e. the subgroup \(U\) of the automorphism group such that:

\[U := \{ s \in G | s(x) - x \in p^{i+1} \text{for all }x\text{ in }M \}\]

If \(F\) is not a normal extension of \({\mathbb{Q}}\) an error will occur.

RamificationGroup(p): RngOrdIdl -> GrpPerm

This is just an abbreviation for RamificationGroup(p, 1).

InertiaGroup(p): RngOrdIdl -> GrpPerm

This is just an abbreviation for RamificationGroup(p, 0).

FixedField(K, U): FldAlg, GrpPerm -> FldNum, Map

Given a normal field \(K\) over \({\mathbb{Q}}\) and a subgroup \(U\) of AutomorphismGroup(K), this returns the largest subfield of \(K\) that is fixed by \(U\).

This function is inverse to FixedGroup.

If \(K\) is not a normal extension of \({\mathbb{Q}}\) an error will occur.

FixedField(K, S): FldAlg, [Map] -> FldAlg, Map

Given a field \(K\) and a sequence of automorphisms of \(K\), this returns the largest subfield of \(K\) that is fixed by the given automorphisms.

FixedGroup(K, L): FldAlg, FldAlg -> GrpPerm

Given a normal field \(K\) over \({\mathbb{Q}}\) and a subfield \(L\), compute the subgroup \(U\) of the AutomorphismGroup(K) that fixes \(L\).

This function is inverse to FixedField.

If \(K\) is not a normal extension of \({\mathbb{Q}}\) an error will occur.

FixedGroup(K, L): FldAlg, [FldAlgElt] -> GrpPerm

Given a normal field \(K\) over \({\mathbb{Q}}\) and a sequence of number field elements \(L\), compute the subgroup \(U\) of the AutomorphismGroup(K) that fixes \(L\).

If \(K\) is not a normal extension of \({\mathbb{Q}}\) an error will occur.

FixedGroup(K, a): FldAlg, FldAlgElt -> GrpPerm

Given a normal field \(K\) over \({\mathbb{Q}}\) and a number field element \(a\), compute the subgroup \(U\) of the AutomorphismGroup(K) that fixes \(a\).

This function is inverse to FixedField.

If \(K\) is not a normal extension of \({\mathbb{Q}}\) an error will occur.

DecompositionField(p): RngOrdIdl -> FldNum, Map

This is an abbreviation for FixedField(K, DecompositionGroup(p)) where \(K\) is the number field of the order of \(p\).

RamificationField(p, i): RngOrdIdl, RngIntElt -> FldNum, Map

This is an abbreviation for FixedField(K, RamificationGroup(p, i)) where \(K\) is the number field of the order of \(p\).

RamificationField(p): RngOrdIdl -> FldNum, Map

This is an abbreviation for FixedField(K, RamificationGroup(p)) where \(K\) is the number field of the order of \(p\).

InertiaField(p): RngOrdIdl -> FldNum, Map

This is an abbreviation for FixedField(K, InertiaField(p)) where \(K\) is the number field of the order of \(p\).

Example: Ramification Groups of a Normal Field (ex-3b0e83)

Run in calculator (opens in a new tab)

We will demonstrate the various groups and fields. In order to do so, we first construct a non-trivial normal field.

> o := MaximalOrder(ext<Rationals()|>.1^4-3);
> os := MaximalOrder(SplittingField(NumberField(o)));
> P := Decomposition(os, 2)[1][1];
> G, M := RayClassGroup(P^3);
> G;
Abelian Group isomorphic to Z/2
Defined on 1 generator
Relations:
    2*G.1 = 0

Since \(G\) is cyclic and the module \(P\) invariant under the automorphisms of \(os\), the class field corresponding to \(G\) will be normal over \(Q\). It Galois group over \(Q\) will be an extension of \(D_4\) by \(C_2\).

> A := AbelianExtension(M);
> O := MaximalOrder(EquationOrder(A));
> Oa := AbsoluteOrder(O);
> Ka := NumberField(Oa);
> Gal, _, Map := AutomorphismGroup(Ka);
> Gal;
Permutation group Gal acting on a set of cardinality 16
Order = 16 = 2^4
    (1, 2, 7, 5)(3, 8, 6, 10)(4, 12, 14, 9)(11, 16, 13, 15)
    (1, 3, 7, 6)(2, 8, 5, 10)(4, 13, 14, 11)(9, 16, 12, 15)
    (1, 4)(2, 9)(3, 11)(5, 12)(6, 13)(7, 14)(8, 15)(10, 16)

Now, let us pick some ideals. The only interesting primes are the primes dividing the discriminant, which in this case will be the primes over \(2\) and \(3\).

> P2 := Decomposition(Oa, 2)[1][1];
> P3 := Decomposition(Oa, 3)[1][1];

First, the valuation of the different of \(Oa\) at \(P2\) should be \(\sum_{i=0}^\infty (\#G(P2, i)-1)\) where \(G(P2, i)\) is the \(i\)-th ramification group.

> s := 0; i := 0;
> repeat
>   G := RamificationGroup(P2, i);
>   s +:= #G-1;
>   print i, "-th ramification group is of order ", #G;
>   i +:= 1;
> until #G eq 1;
0 -th ramification group is of order  8
1 -th ramification group is of order  8
2 -th ramification group is of order  2
3 -th ramification group is of order  2
4 -th ramification group is of order  2
5 -th ramification group is of order  2
6 -th ramification group is of order  1
> s;
18
> Valuation(Different(Oa), P2);
18

According to the theory, \(P2\) should be totally ramified over the inertia field and unramified over \(Q\):

> K2 := InertiaField(P2);
> M2 := MaximalOrder(K2);
> K2r := RelativeField(K2, Ka);
> M2r := MaximalOrder(K2r);
> p2 := M2 meet (MaximalOrder(K2r)!!P2);
> IsInert(p2);
true
> IsTotallyRamified(M2r!!P2);
true

Now we try the same for \(P3\). Since \(3\) is split in \(Ka\), we may consider an additional field: the decomposition field. It should be the maximal subfield if \(K\) such that \(3\) is neither inert (\(f=1\)) nor ramified (\(e=1\)), therefore \(3\) has to split totally.

> D3 := DecompositionField(P3);
> D3M := MaximalOrder(D3);
> IsTotallySplit(3, D3M);
true

The inertia field is the maximal subfield such that \(3\) is unramified. It has to be an extension of \(D3\).

> I3 := InertiaField(P3);
> I3;
Number Field with defining polynomial $.1^4 +
    80346384509631057182412*$.1^3 +
    2256835583037881432653115137736209396615693022*$.\\
    1^2 + 2795818092855476469056989739955845736579291605177\\
    3809455107173769804*$.1 +
    2207787685682553980385342263526644079975418801375161428\\
    41147104301325760481728833650060994 over the Rational
Field
> Discriminant($1);
10700005925626216180895747020647047166414333000723923591882\\
57829873417638072117114945163507537844711544617147344227643\\
21408503489566949866295669400825222748660907808235401444104\\
29329493645714658394673579309893726532999745496689571082958\\
8286937125090034449967033769822464

This (polynomial) discriminant is huge, in fact it is so large that we should avoid the factorisation. We already know the discriminant of \(Ka\). The discriminant of \(I3\) has to be a divisor - so we can use the Discriminant parameter to MaximalOrder: (We are going to need the MaximalOrder for the following embedding.)

> I3M := MaximalOrder(EquationOrder(I3):
> Discriminant := Discriminant(Oa));
> I3M := MaximalOrder(I3);

\(D3\) should be a subfield of \(I3\), so lets verify it:

> IsSubfield(D3, I3);
true Mapping from: FldNum: D3 to FldNum: I3

As a side-effect, Magma is now aware of the embedding and will use it. Without the IsSubfield call, the RelativeField function will fail.

> I3r := RelativeField(D3, I3);
> I3rM := MaximalOrder(I3r);
> K3r := RelativeField(D3, Ka);
> K3rM := MaximalOrder(K3r);
> IsInert(K3rM!!P3 meet D3M, I3rM);
true

The last step: verify that \(P3\) is totally ramified over \(I3\):

> K3r := RelativeField(I3, Ka);
> K3rM := MaximalOrder(K3r);
> IsTotallyRamified(K3rM!!P3 meet I3M, K3rM);
true

Using the decomposition group, we can get the splitting behaviour of any prime in any subfield of \(Ka\).

> L := SubgroupLattice(Gal);
> [ IsNormal(Gal, L[x]) : x in [1..#L]];
[ true, true, true, true, false, false, false, false, true,
true, true, true, true, true, true, false, false, false,
false, true, true, true, true, true, true, true, true ]
> U := L[5];
> k := FixedField(Ka, U);
> kM := MaximalOrder(EquationOrder(k) :
>                    Discriminant := Discriminant(Oa));
> kM := MaximalOrder(k);
> Kr := RelativeField(k, Ka);
> KrM := MaximalOrder(Kr);
> P43 := Decomposition(Oa, 43)[1][1];
> V := DecompositionGroup(P43);

The splitting behaviour is determined by the double coset decomposition of \(Gal\) with respect to \(U\) and \(V\):

> f, I := CosetAction(Gal, U);
> orbs := Orbits(f(V));
> reps := [];
> for o in orbs do
>   _, x := IsConjugate(I, 1, Rep(o));
>   Append(~reps, x @@ f);
> end for;
> reps;
[
    Id(G),
    (1, 2, 7, 5)(3, 8, 6, 10)(4, 12, 14, 9)(11, 16, 13, 15),
    (1, 7)(2, 5)(3, 6)(4, 14)(8, 10)(9, 12)(11, 13)(15, 16),
    (1, 8)(2, 6)(3, 5)(4, 15)(7, 10)(9, 13)(11, 12)(14, 16),
]
> #reps;
4

So there will be at least \(4\) prime ideals over \(43\) in \(k\):

> L := [ ];
> for i in reps do
>   Append(~L, kM meet KrM !! Map(i)(P43));
> end for;
> [ IsPrime(x) : x in L];
[ true, true, true, true ]
> LL := Decomposition(kM, 43);#LL;
4
> [ Position(L, x[1]) : x in LL];
[ 4, 3, 1, 2 ]

Run in calculator (opens in a new tab)

FrobeniusElement(K, p): FldNum, RngIntElt -> GrpPermElt

Compute a Frobenius element at \(p\) in the Galois group of the Galois closure of \(K\). This is a permutation on the roots of a polynomial defining \(K\), which can be recovered as DefiningPolynomial(A) for any Artin representation \(A\) of \(K\); the Frobenius element is well-defined up to conjugacy and modulo inertia.

Example: Frobenius Elements in a Galois Group (ex-87fbb9)

We take a polynomial whose Galois group is \(D_5\) and compute Frobenius elements at \(p=2\) and \(p=5\). They in two different conjugacy classes of 5-cycles in the Galois group.

> load galpols;
> f:=PolynomialWithGaloisGroup(5,2);
> assert IsIsomorphic(GaloisGroup(f),DihedralGroup(5));
> K:=NumberField(f);
> FrobeniusElement(K,2);
(1, 5, 4, 3, 2)
> FrobeniusElement(K,5);
(1, 3, 5, 2, 4)

Run in calculator (opens in a new tab)