%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%
%%  fitting.tex              CRISP documentation           Burkhard H\"ofling
%%
%%  @(#)$Id: fitting.tex,v 1.13 2003/02/11 14:47:19 gap Exp $
%%
%%  Copyright (C) 2000, Burkhard H\"ofling, Mathematisches Institut,
%%  Friedrich Schiller-Universit\"at Jena, Germany
%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\Chapter{Fitting classes and Fitting sets}

In this chapter, you will find information on how to create 
Fitting classes and Fitting sets (see "Creating Fitting classes" and
"Creating Fitting sets" below), and how to compute injectors and radicals
with respect to these;  see "Attributes and operations for Fitting classes
and Fitting sets".


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\Section{Creating Fitting classes}\null

\index{Fitting classes!creating}

Recall that a Fitting class is a nonempty group class which is closed with respect
to normal subgroups and joins of subnormal subgroups.

\>FittingClass(<rec>) O

returns the Fitting class <fitclass> defined by the entries of the record
<rec>. Note that it is the user's responsibility to ensure that <rec> really
defines a Fitting class.
<rec> may have components `\\in', `inj', `rad', `char', and `name'. The
functions assigned to the components are stored in the attributes
`MemberFunction', `InjectorFunction',
`RadicalFunction', `Characteristic', and `Name', of <fitclass>. Please refer
to "MemberFunction",  "InjectorFunction", "RadicalFunction",
"Characteristic", and "ref:Name" for the meaning of these attributes.


 The third example below shows how to construct
the Fitting class $L_2({\cal N})$ (see \cite{DH92}, IX, 1.14 and 1.15), where
$\cal N$ is the class of all finite nilpotent groups.

\beginexample
gap> myNilpotentGroups := FittingClass(rec(\in := IsNilpotent,
>    rad := FittingSubgroup));
FittingClass (in:=<Operation "IsNilpotent">, rad:=<Operation "FittingSubgroup"\
>)
gap> myTwoGroups := FittingClass(rec(
>    \in := G -> IsSubset([2], Set(Factors(Size(G)))), 
>    rad :=  G -> PCore(G,2),
>    inj := G -> SylowSubgroup(G,2)));
FittingClass (in:=function( G ) ... end, rad:=function( G ) ... end, inj:=func\
tion( G ) ... end)
gap> myL2_Nilp := FittingClass (rec (\in := 
>     G -> IsSolvableGroup (G) 
>          and Index (G, Injector (G, myNilpotentGroups)) mod 2 <> 0));
FittingClass (in:=function( G ) ... end)
gap> SymmetricGroup (3) in myL2_Nilp;
false
gap> SymmetricGroup (4) in myL2_Nilp;
true   # thus myL2_Nilp is not closed with respect to factor groups
\endexample

\>FittingProduct(<fit1>, <fit2>) O

returns the Fitting product <prod> of the Fitting classes <fit1> and <fit2>,
i.~e., the class of all groups $G$ such that $G/R$ is  a <fit2>-group, where
$R$ is the <fit1>-radical of $G$.  <prod> is again a Fitting class.
Note that if <fit1> and <fit2> are also
formations, then <prod> equals the
formation product of <fit1> and <fit2>; see "FormationProduct" 
and\indextt{FittingFormationProduct}
"FittingFormationProduct".

\beginexample
gap> FittingProduct (myNilpotentGroups, myTwoGroups);
FittingProduct (FittingClass (in:=<Operation "IsNilpotent">, rad:=<Operation "\
FittingSubgroup">), FittingClass (in:=function( G ) ... end, rad:=function( G \
) ... end, inj:=function( G ) ... end))
gap> FittingProduct (myNilpotentGroups, myL2_Nilp);
FittingProduct (FittingClass (in:=<Operation "IsNilpotent">, rad:=<Operation "\
FittingSubgroup">), FittingClass (in:=function( G ) ... end))
\endexample


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\Section{Creating Fitting formations}\null

\index{Fitting formations!creating}
\index{formations!creating Fitting formations}
\index{Fitting classes!creating Fitting formations}

Fitting formations are Fitting classes which are also formations.

\>FittingFormation(<rec>) O

creates a Fitting formation from the record <rec>. Note that it is the user's 
responsibility to ensure that <rec> really defines a Fitting formation.
<rec> may  have any components admissible for saturated formations (see
"SaturatedFormation")  or Fitting classes (see "FittingClass"), that is,
`\\in', `res', `rad', `inj', `char', and `name', 
 whose values are stored in the attributes `MemberFunction',
`ResidualFunction', `RadicalFunction',
`InjectorFunction', `Characteristic', and `Name', respectively.
Please refer
to "MemberFunction",  "ResidualFunction", "RadicalFunction",
"InjectorFunction",  "Characteristic", and "ref:Name", respectively, for the
meaning of these attributes.


\>SaturatedFittingFormation(<rec>) O

creates a saturated Fitting formation from the record <rec>. Note that it is the user's responsibility to ensure that <rec> really
defines a saturated Fitting formation. <rec> may have any
components admissible for saturated formations (see "SaturatedFormation") or
Fitting classes (see "FittingClass"), that is, `\\in', `res', `proj',
`bound', `locdef', `rad', `inj',   `char', and `Name', whose values are
stored in the attributes `MemberFunction' (see "MemberFunction"),
`ResidualFunction' (see "ResidualFunction"),
 `ProjectorFunction' (see "ProjectorFunction"), `BoundaryFunction' (see
"BoundaryFunction"), `LocalDefinitionFunction' (see
"LocalDefinitionFunction"), `RadicalFunction' (see "RadicalFunction"),
`InjectorFunction' (see "InjectorFunction"), `Characteristic' (see
"Characteristic"), and `Name' (see "ref:Name"), respectively.



%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\Section{Creating Fitting sets}\null

\index{Fitting sets!creating}

A nonempty set $\cal F$ of subgroups of a group $G$ is a *Fitting set of
$G$*  if it satisfies the following properties:

\beginlist
\item{(1)} if $H$ belongs to $\cal F$ and $K$ is normal in $H$, then $K$ 
           belongs to $\cal F$;
         
\item{(2)} if $H$ and $K$ belong to $\cal F$, and $H$ and $K$ are normal in
$\langle H, K \rangle$, then $\langle H, K \rangle = H K$ belongs to $\cal
F$;
         
\item{(3)} if $H$ is in $\cal F$ and $g \in G$, then $H^g$ also belongs to 
           $\cal F$.
\endlist

Note that a Fitting set <fitset> of the group~$G$ is a subset of the set of all
subgroups of~$G$. Therefore it is not closed under group
isomorphisms, hence  is *not* a group class. If $H$ is a subgroup of $G$, then the
subgroups of $G$ in <fitset> which are contained in $H$ form a Fitting set of $H$.
We will not distinguish between <fitset> and the arising Fitting set of $H$. 
Moreover, if  <fit> is a Fitting class and <grp> is a group, then the set of all
subgroups of <grp> which belong to <fit> is a Fitting set of <grp>.

\>IsFittingSet(<G>, <fitset>) O

tests whether <fitset> (or, more precisely, the set of all subgroups of $G$
which are contained in  <fitset>) is a Fitting set of the group <G>. Thus if
<fitset> is a Fitting class, or if
<G> is a subgroup of the group <H> and <fitset> is a Fitting set of <H>,
then `IsFittingSet(<G>, <fitset>)' will return `true'.

\>FittingSet(<G>, <rec>) O

returns the Fitting set <fitset> of the group <G>, defined by the entries of the
record <rec>. Note that, although it would be possible to test whether <rec> 
defines a Fitting set, such a test
is not performed, since it would be extremely expensive, even for relatively
small groups.

<rec> may have components `\\in', `inj', `rad', and `name'. The
functions assigned to the components are stored in the attributes
`MemberFunction', `InjectorFunction', 
`RadicalFunction', and `Name', of <fitset>. Please see "MemberFunction",
"InjectorFunction" and "RadicalFunction" for the meaning of these arguments.

Note that at present, every Fitting set has to be a class (see "Set theoretical
classes"). The second example below shows how to define a Fitting set from 
a list of subgroups.


\beginexample
gap>  fitset := FittingSet(SymmetricGroup (4), rec(
>        \in := S -> IsSubgroup (AlternatingGroup (4), S),
>        rad := S -> Intersection (AlternatingGroup (4), S),
>        inj := S -> Intersection (AlternatingGroup (4), S)));
FittingSet (SymmetricGroup( 
[ 1 .. 4 ] ), rec (in:=function( S ) ... end, rad:=function( S ) ... end, inj:\
=function( S ) ... end))
gap> FittingSet (SymmetricGroup (3), rec(
>       \in := H -> H in [Group (()), Group ((1,2)), Group ((1,3)), Group ((2,3))]));
FittingSet (SymmetricGroup( [ 1 .. 3 ] ), rec (in:=function( H ) ... end))
\endexample

\>ImageFittingSet(<alpha>, <fitset>) O

returns the image <F_1> of the Fitting set <fitset> under the group homomorphism 
<alpha>, i.e.
the Fitting set <F_1> of `Image(<alpha>)' which consists of all subgroups
`<alpha>(<S>)' of `Image(<alpha>)' such that <S> is a <fitset>-injector of 
`PreImage(<alpha>, <S>)'. <fitset> must be a Fitting set of `PreImage(<alpha>)'
or a Fitting class.  Note that the image of a Fitting class is a
Fitting set but not a Fitting class.

\beginexample
gap> alpha := GroupHomomorphismByImages (SymmetricGroup (4), SymmetricGroup (3),
>  [(1,2), (1,3), (1,4)], [(1,2), (1,3), (2,3)]);;
gap> im := ImageFittingSet (alpha, fitset);
FittingSet (Group( [ (1,2), (1,3), (2,3) 
 ] ), rec (inj:=function( G ) ... end))
gap> Radical (Image (alpha), im);
Group([ (1,2,3), (1,3,2) ])
\endexample

\>PreImageFittingSet(<alpha>, <fitset>) O

returns the preimage <fitset_0> of the Fitting set <fitset> of `Image(<alpha>)' under the group homomorphism <alpha>. It consists of all
subgroups <S> of `PreImage(<alpha>)' which are subnormal in `PreImage(<alpha>, <T>)' for some <T> in <fitset>. <fitset> must be a Fitting set of
`Image(<alpha>)' or a Fitting class. 

Note that the preimage of a Fitting class is just a Fitting set but 
not a Fitting class.

Moreover, `ImageFittingSet(PreImageFittingSet(<fitset>, <alpha>),
<alpha>)' equals <fitset> but in general, <fitset> is not contained in
`PreImageFittingSet(ImageFittingSet(<fitset>, <alpha>), <alpha>)'; see
e.g. Example VIII,~2.16 of~\cite{DH92}.

\beginexample
gap> pre := PreImageFittingSet (alpha, NilpotentGroups);
FittingSet (SymmetricGroup( [ 1 .. 4 ] ), rec (inj:=function( G ) ... end))
gap> Injector (Source (alpha), pre);
Group([ (1,4)(2,3), (1,2)(3,4), (2,3,4) ])
\endexample

\>Intersection(<fitset1>, <fitset2>)!{of Fitting sets}

Let <fitset1> and <fitset2> be Fitting sets of the groups <G1> and <G2>. 
Then the intersection of <fitset1> and <fitset2> will be a Fitting set of the 
intersection of <G1> and <G2>. You will run into an error if
{\GAP} cannot compute the intersection of <G1> and <G2>.

\beginexample
gap> F1 := FittingSet (SymmetricGroup (3),
> rec (\in := IsNilpotent, rad := FittingSubgroup));
FittingSet (SymmetricGroup( 
[ 1 .. 3 ] ), rec (in:=<Operation "IsNilpotent">, rad:=<Operation "FittingSubg\
roup">))
gap> F2 := FittingSet (AlternatingGroup (4),
> rec (\in := ReturnTrue, rad := H -> H));
FittingSet (AlternatingGroup( 
[ 1 .. 4 ] ), rec (in:=function(  ) ... end, rad:=function( H ) ... end))
gap> F := Intersection (F1, F2);
FittingSet (Group( 
[ (1,2,3) ] ), rec (in:=function( x ) ... end, rad:=function( G ) ... end))
gap> Intersection (F1, PiGroups ([2,5]));
FittingSet (SymmetricGroup( 
[ 1 .. 3 ] ), rec (in:=function( x ) ... end, rad:=function( G ) ... end))
\endexample


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\Section{Attributes and operations for Fitting classes and Fitting sets}\null

\index{attributes!of Fitting sets}
\index{attributes!of Fitting classes}
\index{operations!for Fitting sets}
\index{operations!for Fitting classes}
\index{Fitting sets!operations for}
\index{Fitting classes!operations for}
\index{Fitting sets!attributes of}
\index{Fitting classes!attributes of}

In addition to operations applicable to classes, both Fitting sets and
Fitting classes admit the following attributes and operations. Of course,
Fitting classes, being group classes, also admit all properties and
attributes for group classes.

\>Radical(<G>,  <fitset>) O

returns the <grpclass>-radical of the group <G>, where
<fitset> is a Fitting set of <G> (see "IsFittingSet"), or a Fitting
class. The <fitset>-radical of <G> is the unique largest normal subgroup
of <G> belonging to <fitset>. Note that `Radical(<G>)' returns the solvable
radical of a group <G> (see "ref:RadicalGroup" in the {\GAP} reference
manual). The class `myL2_Nilp' in the example below has been defined in "FittingClass".

\beginexample
gap> Radical (SymmetricGroup (4), FittingClass (rec(\in := IsNilpotentGroup)));
Group([ (1,4)(2,3), (1,3)(2,4) ])
gap> Radical (SymmetricGroup (4), myL2_Nilp);
Sym( [ 1 .. 4 ] ) 
gap> Radical (SymmetricGroup (3), myL2_Nilp);
Group([ (1,2,3) ])
\endexample

\>Injector(<G>, <fitset>) O

returns a <fitset>-injector of the group <G>, where <fitset> is a
Fitting set of <G> (or a group containing <G>), or a Fitting class. A
subgroup $H$ of $G$ is a <fitset>-injector of <G> if $S \cap H$ is
<fitset>-maximal in $S$ for every subnormal subgroup $S$ of $G$. Note that by 
\cite{DH92}, VIII,~2.9, all <fitset>-injectors of <G> are conjugate in $G$,
and it is not hard to see that every subgroup of <G> has <fitset>-injectors
if and only if <fitset> is a Fitting set of <G>. In particular, if <fitset>
is a group class, then every finite
solvable group has <fitset>-injectors if and only if <fitset> is a Fitting
class; see \cite{DH92}, IX,~1.4.

\beginexample
gap> Injector (SymmetricGroup (4), FittingClass (rec(\in := IsNilpotentGroup)));
Group([ (1,3)(2,4), (1,4)(2,3), (3,4) ])
\endexample

\>RadicalFunction(<class>) A

This attribute, if present, forms part of the definition of <class> supplied
by the user. It must contain a function which takes one argument, a group
$G$, and returns the <class>-radical of $G$. This function
will be used during subsequent calls to `Radical'. Therefore
`Radical' (see "Radical"), which is guaranteed to work for
arbitrary  Fitting sets <class>, should always be called by the user to
compute <class>-radicals.

\>InjectorFunction(<class>) A

This attribute constitutes part of the definition of <class> supplied by the
user. If present, it must contain a function taking a group
$G$ as the only argument and returning a <class>-injector of $G$. This
function will then be used by `Injector' (see "Injector"). Since `Injector'
will work for arbitrary  Fitting sets, it should always be called by the user
to compute <class>-injectors.


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\Section{Functions for minimal normal subgroups and the
socle}\null

This section contains various algorithms to compute the socle of a finite
soluble group, or, more generally, the solvable socle of an arbitrary finite
group. 

\>Socle(<grp>) A

{\CRISP} provides a method for `Socle' (see "ref:Socle") for which works for 
all finite soluble groups <grp>. The socle of a group <grp> is the subgroup 
generated by all minimal normal subgroups of <grp>. See also "SolvableSocle" and 
"PSocle" below.

\beginexample
gap> Size (Socle ( DirectProduct (DihedralGroup (8), CyclicGroup (12))));
12
\endexample

\>AbelianSocle(<grp>) A
\>SolvableSocle(<grp>) A

This function computes the solvable socle of <grp>. The solvable socle of a group <grp> is the
subgroup  generated by all minimal normal solvable subgroups of <grp>.

\>SocleComponents(<grp>) A

This function returns a list of minimal normal subgroups of <grp> such 
that the socle of <grp> (see "Socle") is the direct product of these minimal normal
subgroups. Note that, in general, this decomposition is not unique. Currently,
this function is only implemented for finite soluble groups. See also 
"SolvableSocleComponents" and  "PSocleComponents".

\>AbelianSocleComponents(<grp>) A
\>SolvableSocleComponents(<grp>) A

This function returns a list of solvable minimal normal subgroups of <grp> such 
that the socle of <grp> (see "Socle") is the direct product of these minimal normal
subgroups. Note that, in general, this decomposition is not unique.

\>PSocle(<grp>, <p>) A

If $p$ is a prime, the $p$-socle of a group <grp> is the subgroup 
generated by all minimal normal $p$-subgroups of <grp>.

\>PSocleComponents(<grp>, <p>) A

For a prime $p$, this function returns a list of minimal normal $p$-subgroups of <grp>
such  that the $p$-socle of <grp> (see "PSocle") is the direct product of these minimal normal
subgroups. Note that, in general, this decomposition is not unique.

\>AbelianMinimalNormalSubgroups(<grp>) A

\index{minimal normal subgroups}
This computes a list of all subgroups which are minimal  among the nontrivial
abelian normal subgroups. If <grp> is solvable, this coincides with the list of all 
minimal normal subgroups of <grp>.


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\Section{Low level functions for normal subgroups related to radicals}\null

\index{normal subgroups!with properties inherited by normal subgroups}
\index{invariant normal subgroups!with properties inherited by normal subgroups}


\>OneInvariantSubgroupMaxWrtNProperty(<act>, <grp>, <pretest>, <test>, <data>) O

Let <act> be a list or group whose elements act on <grp> via the caret operator,
such that every subgroup of <grp> invariant under <act> is normal in <grp>. 
Assume $\cal X$ is a set of subgroups of <grp>  such that $\cal X$ contains the trivial group,
and if $M$ and $N$ are <act>-invariant subgroups with $M \in \cal X$ and $M$ containing $N$,
then also $N \in \cal X$. Then 
`OneInvariantSubgroupMaxWrtNProperty' computes an <act>-invariant subgroup $M \in
\cal X$ such that no <act>-invariant subgroup of <grp> properly containing $M$ belongs to $\cal
X$. 

For example, every Fitting set $\cal X$ satisfies the above properties, where $<act> = G$. In
this case, `OneInvariantSubgroupMaxWrtNProperty' will return the $\cal
X$-radical of <grp>.

The class $\cal X$ is described by two functions, <pretest> and <test>. 

<pretest> is a function taking four arguments, <U>, <V>, <R>, and <data>,
where <data> is just the argument passed to
`OneInvariantSubgroupMaxWrtNProperty'. $<U>/<V>$ is an <act>-composition factor of
<grp>, and  <R> is an <act>-invariant subgroup of <grp> contained in <V> which is known to
belong to
$\cal X$.

<pretest> may return the values `true', `false', or `fail'. If it returns 
`true', every <act>-invariant subgroup <N> of <grp> contained in <U> such that
$<N>/<R>$ is 
$G$-isomorphic with $<U>/<V>$ must belong to $\cal X$. If it returns `false',
no such <N> may belong to $\cal X$. 

<test> is a function taking three arguments, <S>, <R>, and <data>, where
data has been described above. <R> is an <act>-invariant subgroup of <grp> belonging to
$\cal X$, and $<S>/<R>$ is an <act>-composition factor of <grp>. The function must return
true if <S> belongs to $\cal X$, and false otherwise. 

Note that `<test>(<S>, <R>, <data>)' is only called if `pretest(<U>, <V>,
<R>, <data>)' has returned `fail' for a chief factor $<U>/<V>$ which is
<G>-isomorphic with $<S>/<R>$. Therefore <test> need not repeat tests
already performed by <pretest>. In particular, if <pretest> always returns
`true' or `false', <test> will not be called at all.

<data> is never used or changed by `OneInvariantSubgroupMaxWrtNProperty', but
exists only as a means for passing additional information to or between
the functions <pretest> and <test>.

\>AllInvariantSubgroupsWithNProperty(<act>, <grp>, <pretest>, <test>, <data>) O

returns a list consisting of all <act>-invariant subgroups of <grp> belonging 
to the class $\cal X$ described by <pretest>, <test>, and <data>. See
the documentation of `OneInvariantSubgroupMaxWrtNProperty' (see
"OneInvariantSubgroupMaxWrtNProperty") for details.

\beginexample
gap> D := DihedralGroup (8);;
gap> AllInvariantSubgroupsWithNProperty (
> D, D,
>     ReturnFail,
>     function (R, S, data)
>         return IsAbelian (R);
>     end,
>     fail);
[ Group([ f3 ]), <pc group with 2 generators>, <pc group with 2 generators>,
  Group([ f2, f3 ]), Group([  ]) ]
\endexample



\>OneNormalSubgroupWithNProperty(<grp>, <pretest>, <test>, <data>) O
\>AllNormalSubgroupsWithNProperty(<grp>, <pretest>, <test>, <data>) O

are the same as `OneInvariantSubgroupMaxWrtNProperty' (see
"OneInvariantSubgroupMaxWrtNProperty")
and  `AllInvariantSubgroupsWithNProperty' ((see
"AllInvariantSubgroupsWithNProperty"), where $<act> = <grp>$, and thus the
<act>-invariant subgroups of <grp> are just the normal subgroups of <grp>.



%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%
%E
%%