############################################################################# ## ## This file is part of GAP, a system for computational discrete algebra. ## This file's authors include Thomas Breuer. ## ## Copyright of GAP belongs to its developers, whose names are too numerous ## to list here. Please refer to the COPYRIGHT file for details. ## ## SPDX-License-Identifier: GPL-2.0-or-later ## ## This file contains methods for fields consisting of cyclotomics. ## ## Note that we must distinguish abelian number fields and fields ## that consist of cyclotomics. ## (The image of the natural embedding of the rational number field ## into a field of rational functions is of course an abelian number field ## but its elements are not cyclotomics since this would be a property given ## by their family.) ##
############################################################################# ## #M IsFiniteDimensional( <A> ) ## ## A finitely generated algebra-with-one that consists of cyclotomics ## has a finite conductor and hence is finite dimensional. ## (Cyclotomic fields and their subfields get the 'IsFiniteDimensional' ## filter from 'AbelianNumberFieldByReducedGaloisStabilizerInfo', ## there is apparently no method for computing the value.) ##
InstallMethod( IsFiniteDimensional, "for an algebra-with-one of cyclotomics",
[ IsAlgebraWithOne and IsCyclotomicCollection
and HasGeneratorsOfAlgebraWithOne ], function( A ) if IsFinite( GeneratorsOfAlgebraWithOne( A ) ) then
return true; fi;
TryNextMethod(); end );
############################################################################# ## #F AbelianNumberFieldByReducedGaloisStabilizerInfo( <F>, <N>, <stab> ) ## ## The constructor `FieldByGenerators' calls this function. ## Since `CyclotomicField' and `AbelianNumberField' generate first the ## information about conductor and Galois stabilizer, it is useful for them ## to call this function instead of constructing generators and calling ## `FieldByGenerators', which would mean to construct <N> and <stab> again. ##
InstallGlobalFunction( AbelianNumberFieldByReducedGaloisStabilizerInfo Source( <Source( id) function( F, N, stab )
local D, d;
D:= Objectify( NewType( CollectionsFamily( CyclotomicsFamily ),
IsField
and IsFiniteDimensional end )
and IsAbelianNumberField
and IsAttributeStoringRep ),
rec() );
d:= Phi(N) / Length( stab );
SetIsCyclotomicField( D, Length( stab ) = 1 );
SetLeftActingDomain( D, F );
SetDegreeOverPrimeField( D, d );
SetGaloisStabilizer( D, stab );
SetConductor( D, N );
SetIsFinite( D, false ); ##
SetDimension( D, d / DegreeOverPrimeField( F ) );
SetPrimeField( D, Rationals );
#M ImageElm( <aut>, <cyc> ) . . for automorphisms of abelian number fields
# Replace generators by their conductor. if not IsInt( xtension ) then
xtension:= fi; if xtension mod 2 = 0 and xtension mod 4 <> 0then
xtension:= xtension / 2; fi;
# The subfield is given by `Rationals' denoting the prime field. if subfield =InstallMethod(ImagesElm,
# The standard field is required. Look whether it is already stored. # If not, generate it and return it
return CYCLOTOMIC_FIELDS( xtension );
elif IsAbelianNumberField( subfield ) then
# CF( subfield, N ) if xtension mod Conductor( subfield ) <> 0then
Error( "<subfield> is not contained in , ], fi;
else
Error( "<subfield> must be `Rationals' or an abelian number field" ); fi;
############################################################################# ## #F ReducedStabilizerInfo( <N>, <stabilizer> ) ## ## is a record with components `N' and `stabilizer', ## which are minimal with the property that they describe the same abelian ## number field as the input parameters <N> and <stabilizer>. ##
BindGlobal( "ReducedGaloisStabilizerInfo", function( N, stabilizer )
local d,
NN,
aut,
pos,
i,
p;
if N#############################################################################
N:= N / 2; fi; if N <= 2then
return rec( N:= 1, stabilizer:= [ 1 ] ); fi;
stabilizer:= Set( #M ImagesSet( <aut>, <field> ) . for automorphisms of abelian number fields
AddSet( stabilizer, 1 );
# Compute the elements of the group generated by `stabilizer'.## for d in stabilizer do
lMethod(ImagesSet,
od;
# reduce the pair `( N, stabilizer )' such that afterwards `N' # describes the conductor of the required field;
gens:= GeneratorsPrimeResidues( N );
NN:= 1; if gens.primes[1] = 2then
if gens.exponents[1] < 3then
notgens.1 in stabilizer then
NN:= NN * 4; fi;
else
# the only case where `gens.generators[i]' is a list; # it contains the generators corresponding to `**' and `*5'; # the first one is irrelevant for the conductor, # except if also the other generator is contained. if gens return F; if not gens.generators[1][1] in stabilizer then
NN:= NN * 4; fi; else
NN:= NN * 4;
aut:= gens.generators[1][2]; while not aut in java.lang.StringIndexOutOfBoundsException: Index 33 out of bounds for length 0
aut:= ( aut ^ 2 ) mod N;
NN:= NN * 2;
od; fi; fi;
pos:= 2; else
pos:= 1; fi;
for i in [ pos .. Length( gens.primes ) ] do
p:= gens.primes[i]; if not gens.generators[i] in stabilizer then
InstallMethod( ImagesRepresentative,
aut:= ( gens.generators[i] ^ ( p - 1 ) ) mod N;
stabilizer do
aut:= ( aut ^ p ) mod N;
NN:= NN * p;
od; fi;
od;
N:= NN; if N <= 2then
stabilizer:= [ 1 ];
N:= 1; else
stabilizer:= Set( stabilizer, x -> x mod N ); fi;
returnrec N: N,stabilizer:= stabilizer ); end );
############################################################################# ## #F AbelianNumberField( <N>, <stab> ) . . . . create an abelian number field ## ## fixed field of the group generated by <stab> (prime residues modulo <N>) ## in the cyclotomic field with conductor <N>. ##
java.lang.StringIndexOutOfBoundsException: Range [4, 2) out of bounds for length 41
local pos, # position in a list
F; # the field, result
# Check the arguments.
Error( "<N> must be a positive integer, <stabilizer> a list" ); fi;
# Compute the conductor and the reduced stabilizer. # Thus the Galois stabilizer component of the field will be minimal.
stabilizer := ReducedGaloisStabilizerInfo(#
N := stabilizer.N;
stabilizer := stabilizer.stabilizer;
if stabilizer = [ 1 ] then
return CyclotomicField( N ); fi;
# The standard field is required. Look whether it is already stored.
return GET_FROM_SORTED_CACHE( java.lang.StringIndexOutOfBoundsException: Index 38 out of bounds for length 27
# Construct the field.
F:= AbelianNumberFieldByReducedGaloisStabilizerInfo "for ANF scalar,
N, stabilizer );
# Return the number field.
return F;
end ); end );
############################################################################# ## #M ViewObj( <F> ) . . . . . . . . . . . . . . view an abelian number field ##
InstallMethod( ViewObj "for abelian number field of cyclotomics",
[ IsAbelianNumberFieldfunction ( aut,elm ) function( F ) if IsPrimeField( LeftActingDomain( F ) ) then(elm, 1 /aut!g)
Print( "NF(", Conductor( F ), ",",
) ) ; else
Print( "AsField( ", LeftActingDomain( F ), " " java.lang.StringIndexOutOfBoundsException: Range [32, 31) out of bounds for length 74 fi; end );
InstallMethod( ViewObj, "for cyclotomic field of cyclotomics",
[ IsCyclotomicField and IsCyclotomicCollection ], function( F ) if IsPrimeField( LeftActingDomain( F ) ) then
Print( "CF(", Conductor( F ), ")" ); else
Print( "AsField( ", LeftActingDomain( F ), ", CF(", Conductor( F ), ") )" ); fi; end );
############################################################################# ## #M PrintObj( <F> ) . . . . . . . . . . . . . . print an abelian number field ##
InstallMethod( PrintObj, "for abelian number field of cyclotomics",
[ IsAbelianNumberField and IsCyclotomicCollection ], function( F ) if IsPrimeField( LeftActingDomain( F ) ) then
[IsFieldHomomorphism and IsANFAutomorphismRep, ,
GaloisStabilizer( F ), ")" ); else
Print( "AsField( ", LeftActingDomain( F ), ", NF(", Conductor( F ), ",", GaloisStabilizer( F ), ") )" ); fi; end );
InstallMethod( PrintObj, "for cyclotomic field of cyclotomics",
[ IsCyclotomicField and IsCyclotomicCollection ], function( F ) if IsPrimeField( LeftActingDomain( F ) ) then
Print( "CF(", Conductor( F ), ")" ); else
Print( "AsField( ", LeftActingDomain( F ), ", CF(", Conductor( F ) return[ GaloisCyc elm,( /aut! java.lang.StringIndexOutOfBoundsException: Index 48 out of bounds for length 48 fi; end );
############################################################################# ## #M String( <F> ) . . . . . . . . . . . . . string of an abelian number field ##
InstallMethod( String, "for abelian number field of cyclotomics",
[ IsAbelianNumberField and IsCyclotomicCollection ], function( F ) if IsPrimeField( LeftActingDomain( F ) ) then
return Concatenation( "NF(",
String( GaloisStabilizer( F ) ), ")" ); else
return Concatenation( "AsField( ", String( LeftActingDomain( F ) ), ##
String( GaloisStabilizer( F ) ), ") )" ); fi; end );
InstallMethod(String, "for cyclotomic field of cyclotomics",
[ IsCyclotomicField and IsCyclotomicCollection ], function( F )
local n;
n:= Conductor( F );
if( (F then if n = 1then
return "Rationals"; elif n = 4then
return "GaussianRationals"; else
return Concatenation( "CF(", String( n ), ")" ); fi; elif n = 4then
return Concatenation( "AsField( ", String( LeftActingDomain( F ) ), ", GaussianRationals )" ); else
return Concatenation( "AsField( ", String( LeftActingDomain( F ) ), ", CF(", String( n ), ") )" ); fi; end );
############################################################################# ## #M \=( <F1>, <F2> ) . . . . . . . . . . comparison of abelian number fields #M \<( <F1>, <F2> ) . . . . . . . . . . comparison of abelian number fields ## ## <F1> is smaller than <F2> if and only if <F1> has a smaller conductor, ## or it has the same conductor as <F2> but its Galois stabilizer component ## is smaller. ##
InstallMethod( \=, "for two abelian number fields",
IsIdenticalObj,
[ IsAbelianNumberField, IsAbelianNumberField ], function ( F1, F2 )
( F1 = Conductor(F2 )
and GaloisStabilizer( F1 ) = GaloisStabilizer( F2 ); end );
InstallMethod( \<, "for two abelian number fields",
IsIdenticalObj,
[, IsAbelianNumberField]java.lang.StringIndexOutOfBoundsException: Index 51 out of bounds for length 51 function ( F1, F2 )
return Conductor( F1 ) < Conductor( F2 )
or ( Conductor( F1 ) = Conductor( F2 )
and GaloisStabilizer( F1 ) < GaloisStabilizer( F2 ) ); end );
############################################################################# ## #M \in( <z>, <F> ) . . . . test if <z> lies in the abelian number field <F> ## ## check whether <z> is a cyclotomic with conductor contained in the ## conductor of <F>, and that <z> is fixed by `GaloisStabilizer( <F> )'. ##
InstallMethod( \in, " cyclotomicand number ",
IsElmsColls,
[ IsCyc, IsAbelianNumberField and IsCyclotomicCollection ], function ( z, F )
return Conductor( F ) mod Conductor( z ) = 0
and ForAll( GaloisStabilizer( F ), x -> GaloisCyc( z, x ) = z ); end );
InstallMethod( \in, "for cyclotomic and cyclotomic field",
IsElmsColls,
[ IsCyc, IsCyclotomicField and IsCyclotomicCollection ], function ( z, F )
return Conductor( F ) mod Conductor( z ) = 0; end );
############################################################################# ## #M Intersection2( <F>, <G> ) . . . . . intersection of abelian number fields ##
InstallMethod( Intersection2,
java.lang.StringIndexOutOfBoundsException: Range [35, 34) out of bounds for length 35
IsIdenticalObj,
[ java.lang.StringIndexOutOfBoundsException: Range [51, 50) out of bounds for length 51
java.lang.StringIndexOutOfBoundsException: Range [27, 23) out of bounds for length 53 function ( F, G )
( Conductor , ); end );
InstallMethod( Intersection2, " cyclotomic and abelian number field"
IsIdenticalObj,
[ IsCyclotomicField and IsCyclotomicCollection,
andIsCyclotomicCollection], function ( F, G )
# intersection of cyclotomic field `F = CF(N)' with number field `G'; # replace `N' by its g.c.d. with the conductor of `G', # and then take the elements of `GaloisStabilizer( G )' modulo `N'. # (If a reduction is necessary, `NF' will do.)
F:= Gcd( Conductor( F ), Conductor( G ) );
return AbelianNumberField( F, Set( GaloisStabilizer( G ),
x -> x mod F ) ); end );
InstallMethod( Intersection2, "for abelian number field and cyclotomic field",
IsIdenticalObj,
[ IsAbelianNumberField and IsCyclotomicCollection,
Fieldand ], function ( G, F )
# intersection of cyclotomic field `F = CF(N)' with number field `G'; # replace `N' by its g.c.d. with the conductor of `G', # and then take the elements of `GaloisStabilizer( G )' modulo `N'. # (If a reduction is necessary, `NF' will do.)
F:= Gcd( Conductor( F ), Conductor( G ) );
return AbelianNumberField( F, Set( GaloisStabilizer( G ),
[ IsFieldHomomorphism and IsANFAutomorphismRep], end );
InstallMethod( Intersection2, " java.lang.StringIndexOutOfBoundsException: Range [13, 12) out of bounds for length 36
IsIdenticalObj,
[ IsAbelianNumberField and IsCyclotomicCollection,
IsAbelianNumberField and IsCyclotomicCollection ], function ( F, G )
local i, j, N, stab, stabF, stabG;
# first compute `N' where `CF(N)' contains the intersection; # reduce the elements of the stabilizers modulo `N', i.e. intersect # `F' and `G' with `CF(N)'; # then compute the corresponding stabilizer, i.e. the product of # stabilizers.
N:= GcdInt( Conductor( F ), Conductor( G ) );
stabF:= Set( GaloisStabilizer( F ), x -> x mod N );
stabG:= Set( java.lang.StringIndexOutOfBoundsException: Index 29 out of bounds for length 2
stab:= []; for i in stabF do for j in stabG do
AddSet( stab, ( i * j ) mod N );
od
od;
# (If a reduction is necessary, `NF' will do.)
AbelianNumberField( ,stab ) end );
############################################################################# ## #M GeneratorsOfDivisionRing( <F> ) . field gens. of an abelian number field #M GeneratorsOfAlgebraWithOne( <F> ) ## ## We have a primitive element of the field extension over the Rationals, ## thus its powers form a basis. ##
Perform( [ GeneratorsOfDivisionRing, GeneratorsOfAlgebraWithOne ], end )java.lang.StringIndexOutOfBoundsException: Index 10 out of bounds for length 10
InstallMethod( op, "for abelian number field of cyclotomics",
[ IsAbelianNumberField and IsCyclotomicCollection ], function( F )
local e;
e:= E( Conductor( F ) );
return [ Sum( GaloisStabilizer( F ), y -> GaloisCyc( e, y ) ) ]; end ); end );
############################################################################# ## #M Conductor( <F> ) . . . . . . . . . conductor of an abelian number field ##
InstallOtherMethod( Conductor, "for abelian number field of cyclotomics",
[ IsAbelianNumberField and IsCyclotomicCollection ],
F -> Conductor( GeneratorsOfField( F ) ) );
############################################################################# ## #M Subfields( <F> ) . . . . . . . . . subfields of an abelian number field ## ## The Galois group of an abelian number field is abelian, ## so the subfields are in bijection with the conjugacy classes of subgroups ## of the Galois group. ##
InstallMethod( Subfields, for field of ,
[ IsAbelianNumberField and IsCyclotomicCollection ], function( F )
local n, stab;
n:= Conductor( F );
stab:= GaloisStabilizer( F );
return Set( ConjugacyClassesSubgroups( GaloisGroup( F ) ),
x -> AbelianNumberField( #############################################################################
List( GeneratorsOfGroup( Representative## #M GaloisGroup( <F> ) . . . . . . . Galois group of an abelian number field end
############################################################################# ## #M PrimeField( <F> ) . . . . . . . . . . . . . . for an abelian number field ##
InstallMethod( PrimeField, "for abelian number field of cyclotomics",
[ IsAbelianNumberField and IsCyclotomicCollection ],
F -> Rationals );
############################################################################# ## #M FieldExtension( <subfield>, <poly> ) . . extend an abelian number field ##
InstallOtherMethod( FieldExtension, "for field of cyclotomics, and univ. polynomial (degree <## If <F> is a field over a proper extension of the rationals then we take #T CollPoly
[ IsAbelianNumberField and IsCyclotomicCollection,
IsLaurentPolynomial ], function( F, poly )
if not IsSubset( F, coeffs ) then
Error( "all coefficients of <poly> must lie in < [ IsAbelianNumberField ], elif3 < Length( coeffs ) then
TryNextMethod(); elif Length( coeffs ) <= 1then
Error( "<poly> must have degree at least 1" ); elif Length( coeffs ) = 2then
# `poly' is a linear polynomial.
root:= - coeffs[1] / coeffs[2];
F:= AsField( F, F );
else
# `poly' has degree 2. The roots of `a*x^2 + b*x + c' are # $\frac{ -b \pm \sqrt{ b^2 - 4ac } }{2a}$.
root:= coeffs[2]^2 - 4 * coeffs[1] * coeffs[3]; if not IsRat( root ) then
java.lang.StringIndexOutOfBoundsException: Index 9 out of bounds for length 0 fi
root:= ( ER( root ) - coeffs[2] ) / ( 2 * coeffs[3] );
F=AsField( F, FieldByGenerators(
Concatenation( GeneratorsOfField( F ), [ root ] ) ) );
fi;
# Store the defining polynomial, and a root of it in the extension field.
SetDefiningPolynomial( F, poly java.lang.StringIndexOutOfBoundsException: Index 0 out of bounds for length 0
SetRootOfDefiningPolynomial( F, root );
"for two abelian number fields of cyclotomics, and cyclotomic",
java.lang.StringIndexOutOfBoundsException: Range [17, 16) out of bounds for length 17
[ IsAbelianNumberField and IsCyclotomicCollection,
IsAbelianNumberField -> 0)java.lang.StringIndexOutOfBoundsException: Index 13 out of bounds for length 13 function( L, K, z )
local N, gal, gens, conj, pnt;
N:= Conductor( L );
# automorphisms of the conductor
gal:= PrimeResidues( N );
if not IsPrimeField( K ) then
# take only the subgroup of `gal' that fixes the subfield pointwise
gens:= GeneratorsOfField( K );
gal:= Filtered( gal,
x -> ForAll( gens, y -> GaloisCyc( y, x ) = y ) ); fi;
# get representatives of cosets of the Galois stabilizer
conj:= [];
gens:= GaloisStabilizer( L ); while gal <> [] do
pnt:= gal[1];
Add( conj, GaloisCyc( z, pnt ) );
SubtractSet( gal, List( gens, x -> ( x * pnt ) mod N ) );
od;
return conj; end );
InstallMethod( Conjugates, "for cycl. field of cyclotomics, ab. number field, and cyclotomic",
IsCollsXElms,
[ IsCyclotomicField and IsCyclotomicCollection,
IsAbelianNumberField and IsCyclotomicCollection, IsCyc ], function( L, K, z )
local conj, Kgens, i;
if not z in L then
Error( "<z> must lie in <L>" ); fi;
if IsPrimeField( K ) then
conj:= List( PrimeResidues( Conductor( L ) ),
i -> GaloisCyc( z, i ) ); else
conj:= [];
Kgens:= GeneratorsOfField( K ); for i in PrimeResidues( Conductor( L ) ) do if ForAll( Kgens, x -> GaloisCyc( x, i ) = x ) then
Add( conj, GaloisCyc( z, i ) ); fi;
od; fi;
return conj; end );
############################################################################# ## #M Norm( <L>, <K>, <z> ) ##
InstallMethod( Norm, "for two abelian number fields of cyclotomics, and cyclotomic",
IsCollsXElms,
[ IsAbelianNumberField and IsCyclotomicCollection,
IsAbelianNumberField and IsCyclotomicCollection, IsCyc ], function( L, K, z )
local N, gal, gens, result, pnt;
N:= Conductor( L );
# automorphisms of the conductor
gal:= PrimeResidues( N );
if not IsPrimeField( K ) then
# take only the subgroup of `gal' that fixes the subfield pointwise
gens:= GeneratorsOfField( K );
gal:= Filtered( gal,
x -> ForAll( gens, y -> GaloisCyc( y, x ) = y ) ); fi;
# get representatives of cosets of `GaloisStabilizer( L )'
result:= 1;
gens:= GaloisStabilizer( L ); while gal <> [] do
pnt:= gal[1];
result:= result * GaloisCyc( z, pnt );
SubtractSet( gal, List( gens, x -> ( x * pnt ) mod N ) );
od;
return result; end );
InstallMethod( Norm, "for cycl. field of cyclotomics, ab. number field, and cyclotomic",
IsCollsXElms,
[ IsCyclotomicField and IsCyclotomicCollection,
IsAbelianNumberField and IsCyclotomicCollection, IsCyc ], function( L, K, z )
local i, result, Kgens;
result:= 1; if IsPrimeField( K ) then for i in PrimeResidues( Conductor( L ) ) do
result:= result * GaloisCyc( z, i );
od; else
Kgens:= GeneratorsOfField( K ); for i in PrimeResidues( Conductor( L ) ) do if ForAll( Kgens, x -> GaloisCyc( x, i ) = x ) then
result:= result * GaloisCyc( z, i ); fi;
od; fi;
return result; end );
############################################################################# ## #M Trace( <L>, K>, <z> ) ##
InstallMethod( Trace, "for two abelian number fields of cyclotomics, and cyclotomic",
IsCollsXElms,
[ IsAbelianNumberField and IsCyclotomicCollection,
IsAbelianNumberField and IsCyclotomicCollection, IsCyc ], function( L, K, z )
local N, gal, gens, result, pnt;
N:= Conductor( L );
# automorphisms of the conductor
gens:= GeneratorsOfField( K );
gal:= PrimeResidues( N );
if not IsPrimeField( K ) then
# take only the subgroup of `gal' that fixes the subfield pointwise
gal:= Filtered( gal,
x -> ForAll( gens, y -> GaloisCyc( y, x ) = y ) ); fi;
# get representatives of cosets of `GaloisStabilizer( L )'
result:= 0;
gens:= GaloisStabilizer( L ); while gal <> [] do
pnt:= gal[1];
result:= result + GaloisCyc( z, pnt );
SubtractSet( gal, List( gens, x -> ( x * pnt ) mod N ) );
od;
return result; end );
InstallMethod( Trace, "for cycl. field of cyclotomics, ab. number field, and cyclotomic",
IsCollsXElms,
[ IsCyclotomicField and IsCyclotomicCollection,
IsAbelianNumberField and IsCyclotomicCollection, IsCyc ], function( L, K, z )
local i, result, Kgens;
result:= 0; if IsPrimeField( K ) then for i in PrimeResidues( Conductor( L ) ) do
result:= result + GaloisCyc( z, i );
od; else
Kgens:= GeneratorsOfField( K ); for i in PrimeResidues( Conductor( L ) ) do if ForAll( Kgens, x -> GaloisCyc( x, i ) = x ) then
result:= result + GaloisCyc( z, i ); fi;
od; fi;
return result; end );
############################################################################# ## #F ZumbroichBase( <n>, <m> ) ## ## returns the set of exponents `e' for which `E(n)^e' belongs to the ## (generalized) Zumbroich base of the cyclotomic field $Q_n$, ## viewed as vector space over $Q_m$. ## ## *Note* that for $n \equiv 2 \bmod 4$ we have ## `ZumbroichBase( <n>, 1 ) = 2 * ZumbroichBase( <n>/2, 1 )' but ## `List( ZumbroichBase( <n>, 1 ), x -> E( <n> )^x ) = ## List( ZumbroichBase( <n>/2, 1 ), x -> E( <n>/2 )^x )'. ##
InstallGlobalFunction( ZumbroichBase, function( n, m )
local nn, base, basefactor, factsn, exponsn, factsm, exponsm, primes,
p, pos, i, k;
if not n mod m = 0then
Error( "<m> must be a divisor of <n>" ); fi;
factsn:= Factors(Integers, n );
primes:= Set( factsn );
exponsn:= List( primes, x -> 0 ); # Product(List( [1..Length(primes)], # x->primes[i]^exponsn[i]))=n
p:= factsn[1];
pos:= 1; for i in factsn do if i > p then
p:= i;
pos:= pos + 1; fi;
exponsn[ pos ]:= exponsn[ pos ] + 1;
od;
factsm:= Factors(Integers, m );
exponsm:= List( primes, x -> 0 ); # Product(List( [1..Length(primes)], # x->primes[i]^exponsm[i]))=m if m <> 1then
p:= factsm[1];
pos:= Position( primes, p ); for i in factsm do if i > p then
p:= i;
pos:= Position( primes, p ); fi;
exponsm[ pos ]:= exponsm[ pos ] + 1;
od; fi;
for k in [ exponsm[i] .. exponsn[i] - 1 ] do
basefactor:= basefactor / primes[i];
base:= Concatenation( List( base, x -> x + basefactor ) );
od;
od;
return Set( base, x -> x mod n ); end );
############################################################################# ## #F LenstraBase( <n>, <stabilizer>, <super>, <m> ) ## ## returns a list of lists of integers; each list indexing the exponents of ## an orbit of a subgroup of <stabilizer> on <n>-th roots of unity. ## ## <super> is a list representing a supergroup of <stabilizer> which ## shall act consistently with the action of <stabilizer>, i.e., each orbit ## of <supergroup> is a union of orbits of <stabilizer>. ## ## ( Shall there be a test if this is possible ? ) ## ## <m> is a positive integer. The basis described by the returned list is ## an integral basis over the cyclotomic field $\Q_m$. ## ## *Note* that the elements are in general not sets, since the first element ## is always an element of `ZumbroichBase( <n>, <m> )'; ## this property is used by `NF' and `Coefficients'. ## ## *Note* that <stabilizer> must not contain the stabilizer of a proper ## cyclotomic subfield of the <n>-th cyclotomic field. ## ## We proceed as follows. ## ## Let $n'$ be the biggest divisor of $n$ coprime to $m$. ## First choose an integral basis $B$ for the extension $\Q_{n'} / \Q$ ## (equivalently, for $\Q_{n} / \Q_{n/n'}$). ## For each element of $B$ choose an integral basis for $\Q_{n/n'} / \Q_m$, ## namely a transversal of $E_m$ in $E_{n/n'}$ where $E_n$ denotes the ## group of $n$-th roots of unity. ## ## The products of elements in these bases form an integral $\Q_m$-basis ## of $\Q_n$. ## Now we choose the bases in such a way that ... ##
InstallGlobalFunction( LenstraBase, function( n, stabilizer, supergroup, m )
local i,
k,
factors, # factors of `n'
primes, # set of prime divisors of `n'
coprimes, # set of prime divisors of `n' coprime to `m'
nprime, # biggest divisor of `n' coprime to `m'
NN, # squarefree part of `n'
zumb, # exponents of roots in the basis of `CF(n)'
N2, # 2-part of `n'
No, # odd part of `n'
transversal, # roots in basis of `CF(n/nprime) / CF(m)', # written as `n'-th roots
orbits,
pnt,
orb,
d,
ppnt,
ord,
a,
neworbits,
rep,
super,
H1;
# We may assume that either `m' is odd or $4$ divides `m'. if m mod 4 = 2then
m:= m / 2; fi;
factors := Factors(Integers, n );
primes := Set( factors );
coprimes := Filtered( primes, x -> m mod x <> 0 );
nprime := Product( Filtered( factors, x -> m mod x <> 0 ) );
NN:= Product( coprimes );
zumb:= List( ZumbroichBase( nprime, 1 ), x -> x * ( n / nprime ) );
transversal := List( ZumbroichBase( n / nprime, m ), x -> x * nprime );
stabilizer:= Set( stabilizer );
orbits:= [];
if nprime = NN then
# $n'$ is squarefree. # We have a normal basis, `stabilizer' acts on `zumb', # we do not consider equivalence classes since they are all trivial, # and `supergroup' is obsolete since `zumb' describes a normal basis.
# *Note* that if $n'$ is even then `zumb' does not consist of # at least `NN'-th roots!
while0 < Length( zumb ) do
# Compute the orbit of `stabilizer' of a point in `zumb'.
pnt:= zumb[1];
# For each root in `transversal', compute the orbit of `n'-th roots # under `stabilizer'.
neworbits:= List( transversal,
root -> List( stabilizer,
x -> ( root + pnt ) * x mod n ) );
SubtractSet( zumb, neworbits[1] );
Append( orbits, neworbits );
od;
else
# Let $d(i)$ be the largest squarefree number whose square divides the # order of $e_{n'}^i$, that is $n' / \gcd( n', i )$. # Define an equivalence relation on the set $S$ of at least `NN'-th # roots of unity. # $i$ and $j$ are equivalent iff $n'$ divides $( i - j ) d(i)$. The # equivalence class $(i)$ of $i$ is # $\{ i + k n' / d(i) ; 0 \leq k \< d(i) \}$.
# For the case that `NN' is even, replace those roots in $S$ with order # not divisible by 4 by their negatives. # (Equivalently\: Replace *all* elements in $S$ by their negatives.)
# If 8 does not divide $n'$ and $n' \not= 4$, `zumb' is a subset of $S$, # the intersection of $(i)$ with `zumb' is of order $\varphi( d(i) )$, # it is a basis for the $Z$--submodule spanned by $(i)$. # Furthermore, the minimality of `n' yields that `stabilizer' acts fixed # point freely on the set of equivalence classes.
# More exactly, fixed points occur exactly if there is an element `s' in # `stabilizer' which is congruent $-1$ modulo `N2' and congruent $+1$ # modulo `No'.
# The base is constructed as follows\: # # Until all classes are touched: # 1. Take a point `pnt' (in `zumb'). # 2. Choose a maximal linear independent set `pnts' in the equivalence # class of `pnt' (the intersection of the class with `zumb'). # 3. Take the `stabilizer'--orbits of `pnts' as base elements; # remove the touched equivalence classes. # 4. For the representatives `rep' in `supergroup'\: # If `rep' maps `pnt' to an equivalence class that was not yet # touched, take the `stabilizer'--orbits of the images of `pnts' # under `rep' as base elements; # remove the touched equivalence classes.
# Compute nontriv. representatives of `supergroup' over `stabilizer'.
super:= Difference( supergroup, stabilizer );
supergroup:= []; while0 < Length( super ) do
pnt:= super[1];
Add( supergroup, pnt );
SubtractSet( super, List( stabilizer, x -> ( x * pnt ) mod n ) );
od;
# Compute 2-part and odd part of $n'$.
N2 := 1;
No := nprime; while No mod 2 = 0do
N2:= N2 * 2;
No:= No / 2;
od;
# Compute the subgroup `H1' of `stabilizer' that acts fixed point # freely on the set of equivalence classes, # and the element `a' that (if exists) fixes some classes pointwise.
H1 := [];
a := 0; for k in stabilizer do if k mod 4 = 1then
Add( H1, k ); elif ( k -1 ) mod No = 0
and ( ( k + 1 ) mod N2 = 0 or ( k + 1 - N2/2 ) mod N2 = 0 ) then
a:= k; fi;
od; if a = 0then
H1:= stabilizer; fi;
while0 < Length( zumb ) do
neworbits:= [];
pnt:= zumb[1];
d:= 1;
ord:= n / GcdInt( n, pnt ); for i in coprimes do if ord mod i^2 = 0then d:= d * i; fi;
od;
if ( a = 0 ) or ( ord mod 8 = 0 ) then
# No `H1'-orbit can be fixed by `a'.
for k in [ 0 .. d-1 ] do
# Loop over the equivalence class of `pnt', # consider only the points in `zumb'.
ppnt:= pnt + k * n / d; if ppnt in zumb then
orb:= List( stabilizer, x -> ( ppnt * x ) mod n );
Append( neworbits,
List( transversal,
root -> List( stabilizer,
x -> ( root + ppnt ) * x mod n ) ) );
fi;
od;
elif ord mod 4 = 0then
# `a' maps each point in the orbit of `H1' to its inverse, # we ignore all these points.
orb:= List( stabilizer, x -> ( pnt * x ) mod n );
else
# The orbit of `H1' is pointwise fixed by `a'. for k in [ 0 .. d-1 ] do
ppnt:= pnt + k * n / d; if ppnt in zumb then
orb:= List( H1, x -> ( ppnt * x ) mod n );
Append( neworbits,
List( transversal,
root -> List( H1,
x -> ( root + ppnt ) * x mod n ) ) );
fi;
od;
fi;
# Remove the equivalence classes of all new points from `zumb'. for pnt in orb do
SubtractSet( zumb, List( [ 0 .. d-1 ],
k -> ( pnt + k * n / d ) mod n ) );
od;
Append( orbits, neworbits );
# use `supergroup'\: # Is there a point in `zumb' that is not equivalent to # `( pnt * rep ) mod nprime' ? # (Note that the factor group `supergroup / stabilizer' acts on the # set of unions of orbits with equivalent elements.)
for rep in supergroup do
# is there an `x' in `zumb' that is equivalent to `pnt * rep' ? if ForAny( zumb, x -> ( ( x - pnt * rep ) * d ) mod n = 0 ) then
Append( orbits, List( neworbits,
x -> List( x, y -> (y*rep) mod n ) ) ); for ppnt in Last(orbits) do
SubtractSet( zumb, List( [ 0..d-1 ],
k -> ( ppnt + k * n / d ) mod n ) );
od; fi;
od;
od; fi;
# Return the list of orbits.
return orbits; end );
############################################################################# ## #R IsCanonicalBasisAbelianNumberFieldRep( <B> ) ## ## The canonical basis of a number field is defined to be a Lenstra basis ## in the case that the subfield is `Rationals'. #T extend this to the case where the subfield is a cyclotomic field! ## In all other cases a normal basis is chosen. ##
DeclareRepresentation( "IsCanonicalBasisAbelianNumberFieldRep",
IsAttributeStoringRep,
[ "coeffslist", "coeffsmat", "lenstrabase", "conductor" ] );
############################################################################# ## #R IsCanonicalBasisCyclotomicFieldRep( <B> ) ## ## The canonical basis of a field extension $F / K$ for a cyclotomic field ## $F$ is the Zumbroich basis if $K$ is a cyclotomic field. ## Otherwise it is a normal basis. ##
DeclareRepresentation( "IsCanonicalBasisCyclotomicFieldRep",
IsCanonicalBasisAbelianNumberFieldRep,
[ "zumbroichbase" ] );
############################################################################# ## #M CanonicalBasis( <F> ) ## ## The canonical basis of a number field is defined to be a Lenstra basis ## in the case that the subfield is a cyclotomic field. ## ## In all other cases a normal basis is chosen. ##
InstallMethod( CanonicalBasis, "for abelian number field of cyclotomics",
[ IsAbelianNumberField and IsCyclotomicCollection ], function ( F )
local N, # conductor of `F'
k,
lenst,
i,
B,
BB,
normalbase,
subbase,
m,
j,
C,
coeffsmat,
val,
l;
# Make the basis object.
B:= Objectify( NewType( FamilyObj( F ),
IsFiniteBasisDefault
and IsCanonicalBasis
and IsCanonicalBasisAbelianNumberFieldRep ),
rec() );
SetUnderlyingLeftModule( B, F );
if IsCyclotomicField( LeftActingDomain( F ) ) then
# Compute the standard Lenstra basis and the `coeffslist' component. # If `GaloisStabilizer( F )' acts fixed point freely on the # equivalence classes we must change from the Zumbroich basis to a # `GaloisStabilizer( F )'-normal basis, # and afterwards choose coefficients with respect to that basis. # In the case of fixed points, only the subgroup `H1' of index 2 in # `GaloisStabilizer( F )' acts fixed point freely; # we change to a `H1'-normal basis, and afterwards choose coefficients.
# For this basis <B> we want a component `coeffslist' such that # in the special case of a field over the rationals we have # `CoeffsCyc( z, N ){ <B>!.coeffslist } = Coefficients( <B>, z )'.
N:= Conductor( F );
lenst:= LenstraBase( N, GaloisStabilizer( F ), GaloisStabilizer( F ),
Conductor( LeftActingDomain( F ) ) );
# Fill in additional components.
SetBasisVectors( B, List( lenst,
x -> Sum( List( x, y -> E(N)^y ) ) ) );
B!.coeffslist := MakeImmutable(List( lenst, x -> x[1] + 1 ));
B!.lenstrabase := MakeImmutable(lenst);
B!.conductor := N; #T better compute basis vectors only if necessary #T (in the case of a normal basis the vectors are of course known ...)
SetIsIntegralBasis( B, true );
else
# A basis of an extension of a number field is a normal basis. #T handle extensions of cycl. fields specially!! # Coefficients can be computed using a tensor form basis for that # the base change relative to the Lenstra basis (relative to the # rationals) is computed.
# Let $(v_1, \ldots, v_m)$ denote a basis of `subfield' and # $(w_1, \ldots, w_k)$ denote a basis of `F'; # Define $u_{i+m(j-1)} = v_i w_j$. Then $(u_l; 1\leq l\leq mk)$ # is a basis of `F' over the rationals. # First change from the Lenstra basis to this basis; the matrix is `C'.
normalbase:= NormalBase( F );
subbase:= BasisVectors( CanonicalBasis( LeftActingDomain( F ) ) );
BB:= CanonicalBasis( AbelianNumberField( Conductor( F ),
GaloisStabilizer( F ) ) );
m:= Length( subbase );
k:= Length( normalbase );
N:= Conductor( normalbase );
C:= []; for j in normalbase do
# Compute the Lenstra basis coefficients. for i in subbase do
Add( C, Coefficients( BB, i*j ) );
od;
od;
C:= C^(-1);
# Let $(c_1, \ldots, c_{mk})$ denote the coefficients with respect # to the new base. To achieve `<coeffs> \* normalbase = <z>' we have # to take $\sum_{i=1}^m c_{i+m(j-1)} v_i$ as $j$--th coefficient:
coeffsmat:= []; for i in [ 1 .. Length( C ) ] do# for all rows
coeffsmat[i]:= []; for j in [ 1 .. k ] do
val:= 0; for l in [ 1 .. m ] do
val:= val + C[i][ m*(j-1)+l ] * subbase[l];
od;
coeffsmat[i][j]:= val;
od;
od;
# Multiplication of a Lenstra basis coefficient vector with # `coeffsmat' means first changing to the base of products # $v_i w_j$ and then summation over the parts of the $v_i$.
############################################################################# ## #M Basis( <F> ) ##
InstallMethod( Basis, "for abelian number field of cyclotomics (delegate to `CanonicalBasis')",
[ IsAbelianNumberField and IsCyclotomicCollection ],
CANONICAL_BASIS_FLAGS,
CanonicalBasis );
############################################################################# ## #M Coefficients( <B>, <z> ) . . . . . for canon. basis of ab. number field ##
InstallMethod( Coefficients, "for canonical basis of abelian number field, and cyclotomic",
IsCollsElms,
[ IsBasis and IsCanonicalBasis and IsCanonicalBasisAbelianNumberFieldRep,
IsCyc ], function ( B, z )
local V,
F,
coeffs,
n,
m,
zumb,
NN,
Em;
if not z in UnderlyingLeftModule( B ) then
return fail; fi;
V:= UnderlyingLeftModule( B );
F:= LeftActingDomain( V );
# The information about the standard Lenstra basis coefficients # is stored in `B!.coeffslist'. if IsPrimeField( F ) then
# Take the relevant sublist, this suffices for extensions # of the rationals.
coeffs:= CoeffsCyc( z, B!.conductor ){ B!.coeffslist };
elif IsCyclotomicField( F ) then
# `B' is an integral basis of an extension of a cyclotomic field $\Q_m$, # the coefficient of the root $\zeta$ is # $\sum_{\eta\in\B_m} a_{\eta\zeta} \eta$.
coeffs:= CoeffsCyc( z, B!.conductor );
n:= Conductor( V );
m:= Conductor( F );
zumb:= CanonicalBasis( F )!.zumbroichbase;
NN:= n/m;
Em:= E(m);
coeffs:= List( B!.coeffslist,
j->Sum( zumb, k->coeffs[ ((k*NN+j-1) mod n )+1 ]*Em^k ) );
fi;
if IsBound( B!.coeffsmat ) then
# Compute the coefficients with respect to the field extension.
coeffs:= CoeffsCyc( z, B!.conductor ){ B!.coeffslist } * B!.coeffsmat;
fi;
# Return the coefficients list.
return coeffs; end );
# Handle trivial cases. if N = 1then
return Rationals; elif N = 4then
return GaussianRationals; fi;
# Compute the reduced stabilizer info.
stab:= Filtered( PrimeResidues( N ),
x -> ForAll( gens,
gen -> GaloisCyc( gen, x ) = gen ) );
# Construct and return the field.
return AbelianNumberFieldByReducedGaloisStabilizerInfo( Rationals, N,
stab ); end );
InstallMethod( FieldByGenerators, "for field and collection, both collections of cyclotomics",
IsIdenticalObj,
[ IsField and IsCyclotomicCollection, IsCyclotomicCollection ], function( F, gens )
local N, stab;
N:= Conductor( gens );
if F = Rationals then
# Handle trivial cases. if N = 1then
return Rationals; elif N = 4then
return GaussianRationals; fi;
else
N:= Lcm( N, Conductor( F ) );
gens:= Concatenation( gens, GeneratorsOfField( F ) ); fi;
# Compute the reduced stabilizer info.
stab:= Filtered( PrimeResidues( N ),
x -> ForAll( gens,
gen -> GaloisCyc( gen, x ) = gen ) );
# Construct and return the field.
return AbelianNumberFieldByReducedGaloisStabilizerInfo( F, N, stab ); end );
############################################################################# ## #M CanonicalBasis( <F> ) ## ## The canonical basis of a field extension $F / K$ for a cyclotomic field ## $F$ is the Zumbroich basis if $K$ is a cyclotomic field. ## Otherwise it is the first normal basis. ##
InstallMethod( CanonicalBasis, "for cyclotomic field of cyclotomics",
[ IsCyclotomicField and IsCyclotomicCollection ], function( F )
local n,
m,
B,
subfield,
zumb,
subvectors,
vectors,
i, j, k, l,
C,
coeffsmat,
val;
n:= Conductor( F );
B:= Objectify( NewType( FamilyObj( F ),
IsFiniteBasisDefault
and IsCanonicalBasis
and IsCanonicalBasisCyclotomicFieldRep ),
rec() );
SetUnderlyingLeftModule( B, F );
B!.conductor:= n;
subfield:= LeftActingDomain( F );
if IsCyclotomicField( subfield ) then
SetIsIntegralBasis( B, true );
# Construct the Zumbroich basis.
B!.zumbroichbase := MakeImmutable(ZumbroichBase( n, Conductor( subfield ) ));
else
# Compute a normal basis.
# Let $(v_1, \ldots, v_m)$ denote `Basis( F.field ).vectors' and # $(w_1, \ldots, w_k)$ denote `vectors'. # Define $u_{i+m(j-1)} = v_i w_j$. Then $(u_l; 1\leq l\leq mk)$ # is a $Q$-basis of `F'. First change from the Zumbroich basis to # this basis; the matrix is `C'\:
zumb := ZumbroichBase( n, 1 ) + 1;
subvectors := BasisVectors( Basis( subfield ) );
vectors := NormalBase( F );
m:= Length( subvectors );
k:= Length( vectors );
C:= []; for j in vectors do for i in subvectors do
Add( C, CoeffsCyc( i*j, n ){ zumb } );
od;
od;
C:= C^(-1);
# Let $(c_1, \ldots, c_{mk})$ denote the coefficients with respect # to the new basis. # To achieve `<coeffs> \* BasisVectors( <B> ) = <z>' we have # to take $\sum_{i=1}^m c_{i+m(j-1)} v_i$ as $j$--th coefficient\:
coeffsmat:= []; for i in [ 1 .. Length( C ) ] do# for all rows
coeffsmat[i]:= []; for j in [ 1 .. k ] do
val:= 0; for l in [ 1 .. m ] do
val:= val + C[i][ m*(j-1)+l ] * subvectors[l];
od;
coeffsmat[i][j]:= val;
od;
od;
############################################################################# ## #M BasisVectors( <B> ) ##
InstallMethod( BasisVectors, "for canon. basis of cyclotomic field of cyclotomics",
[ IsBasis and IsCanonicalBasis and IsCanonicalBasisCyclotomicFieldRep ], function( B )
local e; # Basis vectors are bound if the subfield is not a cycl. field. #T ??
e:= E( Conductor( UnderlyingLeftModule( B ) ) );
return List( B!.zumbroichbase, x -> e^x ); end );
############################################################################# ## #M Coefficients( <B>, <z> ) . . . . . . . . for canon. basis of cycl. field ##
InstallMethod( Coefficients, "for canonical basis of cyclotomic field, and cyclotomic",
IsCollsElms,
[ IsBasis and IsCanonicalBasis and IsCanonicalBasisCyclotomicFieldRep,
IsCyc ], function( B, z )
local N,
coeffs,
F,
m,
zumb,
NN,
Em;
F:= UnderlyingLeftModule( B ); if not z in F then return fail; fi;
N:= B!.conductor;
# Get the Zumbroich basis representation of <z> in `N'-th roots.
coeffs:= CoeffsCyc( z, N ); if coeffs = fail then return fail; fi;
F:= LeftActingDomain( F );
if IsPrimeField( F ) then
# Get the Zumbroich basis coefficients (basis $B_{n,1}$)
coeffs:= coeffs{ B!.zumbroichbase + 1 };
elif IsCyclotomicField( F ) then
# Get the Zumbroich basis coefficients (basis $B_{n,m}$) directly.
m:= Conductor( F );
zumb:= CanonicalBasis( F )!.zumbroichbase;
NN:= N/m;
Em:= E(m);
coeffs:= List( B!.zumbroichbase,
j->Sum( zumb, k->coeffs[ ((k*NN+j) mod N )+1 ]*Em^k ) );
else
# The subfield is not a cyclotomic field. # The necessary information is stored in `B!.coeffsmat'.
coeffs:= coeffs{ B!.zumbroichbase + 1 } * B!.coeffsmat;
fi;
# Return the list of coefficients.
return coeffs; end );
############################################################################# ## ## Automorphisms of abelian number fields ##
InstallMethod( ExponentOfPowering, "for an identity mapping",
[ IsMapping and IsOne ],
map -> 1 );
InstallMethod( ExponentOfPowering, "for a mapping (check whether it is the identity mapping)",
[ IsMapping ], function( map ) if IsOne( map ) then
return 1; else
TryNextMethod(); fi; end );
############################################################################# ## #F ANFAutomorphism( <F>, <k> ) . . automorphism of an abelian number field ##
InstallGlobalFunction( ANFAutomorphism, function ( F, k )
local galois, aut;
# check the arguments if not ( IsAbelianNumberField(F) and IsCyclotomicCollection(F) ) then
Error("<F> must be an abelian number field consisting of cyclotomics"); elif not IsRat( k ) then
Error( "<k> must be an integer" ); fi; if not IsInt( k ) then #T this is a hack ...
k:= k mod Conductor( F ); fi; if Gcd( Conductor( F ), k ) <> 1then
Error( "<k> must be coprime to the conductor of <F>" ); fi;
# Let $F / K$ be a field extension where $Q_n$ is the conductor of $F$; # let $S(F)$ be the group of those prime residues mod $n$ that fix $F$ # pointwise. The Galois group of $F / K$ is in natural correspondence # to $S(K) / S(F)$. Thus each automorphism of $F / K$ corresponds to # a coset $c$ of $S(K)$, and it acts on $F$ like each element of $c$. # The automorphism `ANFAutomorphism( F/K, k )' maps $x\in F / K$ to # `GaloisCyc( <x>, k )'.
# Choose the smallest representative ...
galois:= Set(GaloisStabilizer( F ), x->x*k mod Conductor( F ))[1]; if galois = 1then
return IdentityMapping( F ); fi;
# make the mapping
aut:= Objectify( TypeOfDefaultGeneralMapping( F, F,
IsSPGeneralMapping
and IsANFAutomorphism ),
rec() );
aut!.galois:= galois;
return aut; end );
############################################################################# ## #M \=( <aut1>, <aut2> ) . . . . for automorphisms of abelian number fields #M \=( <id>, <aut> ) . . . . . . for automorphisms of abelian number fields #M \=( <aut>, <id> ) . . . . . . for automorphisms of abelian number fields ##
InstallMethod( \=, "for two ANF automorphisms",
IsIdenticalObj,
[ IsFieldHomomorphism and IsANFAutomorphismRep,
IsFieldHomomorphism and IsANFAutomorphismRep ], function ( aut1, aut2 )
return Source( aut1 ) = Source( aut2 )
and aut1!.galois = aut2!.galois; end );
InstallMethod( \=, "for identity mapping and ANF automorphism",
IsIdenticalObj,
[ IsMapping and IsOne,
IsFieldHomomorphism and IsANFAutomorphismRep ], function ( id, aut )
return Source( id ) = Source( aut )
and aut!.galois = 1; end );
InstallMethod( \=, "for ANF automorphism and identity mapping",
IsIdenticalObj,
[ IsFieldHomomorphism and IsANFAutomorphismRep,
IsMapping and IsOne ], function ( aut, id )
return Source( id ) = Source( aut )
and aut!.galois = 1; end );
############################################################################# ## #M \<( <aut1>, <aut2> ) . . . . for automorphisms of abelian number fields ##
InstallMethod( \<, "for two ANF automorphisms",
IsIdenticalObj,
[ IsFieldHomomorphism and IsANFAutomorphismRep,
IsFieldHomomorphism and IsANFAutomorphismRep ], function ( aut1, aut2 )
return Source( aut1 ) < Source( aut2 )
or ( Source( aut1 ) = Source( aut2 )
and aut1!.galois < aut2!.galois ); end );
InstallMethod( \<, "for identity mapping and ANF automorphism",
IsIdenticalObj,
[ IsMapping and IsOne,
IsFieldHomomorphism and IsANFAutomorphismRep ], function ( id, aut )
return Source( id ) < Source( aut )
or ( Source( id ) = Source( aut )
and 1 < aut!.galois ); end );
InstallMethod( \<, "for ANF automorphism and identity mapping",
IsIdenticalObj,
[ IsFieldHomomorphism and IsANFAutomorphismRep,
IsMapping and IsOne ], function ( aut, id )
return Source( aut ) < Source( id ); end );
############################################################################# ## #M ImageElm( <aut>, <cyc> ) . . for automorphisms of abelian number fields ##
InstallMethod( ImageElm, "for ANF automorphism and scalar",
FamSourceEqFamElm,
[ IsFieldHomomorphism and IsANFAutomorphismRep, IsCyc ], function ( aut, elm )
return GaloisCyc( elm, aut!.galois ); end );
############################################################################# ## #M ImagesElm( <aut>, <cyc> ) . . for automorphisms of abelian number fields ##
InstallMethod( ImagesElm, "for ANF automorphism and scalar",
FamSourceEqFamElm,
[ IsFieldHomomorphism and IsANFAutomorphismRep, IsScalar ], function ( aut, elm )
return [ GaloisCyc( elm, aut!.galois ) ]; end );
############################################################################# ## #M ImagesSet( <aut>, <field> ) . for automorphisms of abelian number fields ##
InstallMethod( ImagesSet, "for ANF automorphism and field",
CollFamSourceEqFamElms,
[ IsFieldHomomorphism and IsANFAutomorphismRep, IsField ], function ( aut, F )
return F; end );
############################################################################# ## #M ImagesRepresentative( <aut>, <cyc> ) . . for autom. of ab. number fields ##
InstallMethod( ImagesRepresentative, "for ANF automorphism and scalar",
FamSourceEqFamElm,
[ IsFieldHomomorphism and IsANFAutomorphismRep, IsScalar ], function ( aut, elm )
return GaloisCyc( elm, aut!.galois ); end );
############################################################################# ## #M PreImageElm( <aut>, <cyc> ) . . . . . . . for autom. of ab. number fields ##
InstallMethod( PreImageElm, "for ANF automorphism and scalar",
FamRangeEqFamElm,
[ IsFieldHomomorphism and IsBijective and IsANFAutomorphismRep,
IsScalar ], function ( aut, elm )
return GaloisCyc( elm, ( 1 / aut!.galois )
mod Conductor( Range( aut ) ) ); end );
############################################################################# ## #M PreImagesElm( <aut>, <cyc> ) . . . . . . for autom. of ab. number fields ##
InstallMethod( PreImagesElm, "for ANF automorphism and scalar",
FamRangeEqFamElm,
[ IsFieldHomomorphism and IsANFAutomorphismRep, IsScalar ], function ( aut, elm )
return [ GaloisCyc( elm, ( 1 / aut!.galois )
mod Conductor( Range( aut ) ) ) ]; end );
############################################################################# ## #M PreImagesSet( <aut>, <field> ) . . . . . for autom. of ab. number fields ##
InstallMethod( PreImagesSet, "for ANF automorphism and scalar",
CollFamRangeEqFamElms,
[ IsFieldHomomorphism and IsANFAutomorphismRep, IsField ], function ( aut, F )
return F; end );
############################################################################# ## #M PreImagesRepresentative( <aut>, <cyc> ) . for autom. of ab. number fields ##
InstallMethod( PreImagesRepresentative, "for ANF automorphism and scalar",
FamRangeEqFamElm,
[ IsFieldHomomorphism and IsANFAutomorphismRep, IsScalar ], function ( aut, elm )
return GaloisCyc( elm, ( 1 / aut!.galois )
mod Conductor( Range( aut ) ) ); end );
############################################################################# ## #M CompositionMapping2( <aut2>, <aut1> ) . . for autom. of ab. number fields ##
InstallMethod( CompositionMapping2, "for two ANF automorphisms",
FamSource1EqFamRange2,
[ IsFieldHomomorphism and IsANFAutomorphismRep,
IsFieldHomomorphism and IsANFAutomorphismRep ], function ( aut1, aut2 )
return ANFAutomorphism( Source( aut1 ), aut1!.galois * aut2!.galois ); end );
############################################################################# ## #M InverseGeneralMapping( <aut> ) . . . . . for autom. of ab. number fields ##
InstallOtherMethod( InverseGeneralMapping, "for ANF automorphism",
[ IsFieldHomomorphism and IsANFAutomorphismRep ],
aut -> ANFAutomorphism( Source( aut ), 1 / aut!.galois ) );
############################################################################# ## #M \^( <aut>, <n> ) . . . . . . . . . . . . for autom. of ab. number fields ##
InstallMethod( \^, "for ANF automorphism and integer",
[ IsFieldHomomorphism and IsANFAutomorphismRep, IsInt ], function ( aut, i )
return ANFAutomorphism( Source( aut ), aut!.galois^i ); end );
############################################################################# ## #M PrintObj( <aut> ) . . . . . . . . . . . . for autom. of ab. number fields ##
InstallMethod( PrintObj, "for ANF automorphism",
[ IsFieldHomomorphism and IsANFAutomorphismRep ], function ( aut )
Print( "ANFAutomorphism( ", Source( aut ), ", ", aut!.galois, " )" ); end );
############################################################################# ## #M GaloisGroup( <F> ) . . . . . . . Galois group of an abelian number field ## ## The required group is a factor group of the Galois group $G$ ## of the enveloping cyclotomic field. ## So the group $U$ generated by the actions of the generators of $G$ on <F> ## is the Galois group of <F>, viewed as field over the rationals. ## ## If <F> is a field over a proper extension of the rationals then we take ## the pointwise stabilizer of the subfield in $U$. ##
InstallMethod( GaloisGroup, "for abelian number field ",
[ IsAbelianNumberField ], function( F )
local group;
group:= GroupByGenerators( List( Flat(
GeneratorsPrimeResidues( Conductor( F ) ).generators ),
x -> ANFAutomorphism( F, x ) ),
IdentityMapping( F ) );
if not IsPrimeField( LeftActingDomain( F ) ) then
group:= Stabilizer( group,
GeneratorsOfField( LeftActingDomain( F ) ), OnTuples ); fi;
return group; end );
InstallMethod( Representative,
[ IsAdditiveMagmaWithZero and IsCyclotomicCollection ],
f -> 0 );
Die Informationen auf dieser Webseite wurden
nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit,
noch Qualität der bereit gestellten Informationen zugesichert.
Bemerkung:
Die farbliche Syntaxdarstellung und die Messung sind noch experimentell.