Elementary and hyperelementary groups #
Fix a natural number p. A group is p-elementary when it is the internal direct product of a
cyclic subgroup of order prime to p and a p-subgroup, and p-hyperelementary when it has a
cyclic normal subgroup of order prime to p whose quotient is a p-group; a group is
elementary (respectively hyperelementary) when it is p-elementary (respectively
p-hyperelementary) for some prime p. Mathlib has IsCyclic and IsPGroup but no predicate
built from them in this shape, so this file supplies the two definitions together with the closure
properties they are used through.
The definitions are the ones Brauer's induction theorem is stated against: Artin's theorem writes a
character as a rational combination of characters induced from cyclic subgroups, and Brauer's
sharpens this to an integral combination of characters induced from elementary subgroups. That
the elementary groups include the cyclic ones is the content of
TauCeti.isPElementary_of_isCyclic below: a finite cyclic group is p-elementary for every
prime p, by splitting it as the product of its p-part and its p'-part.
The second half of the file builds the p-elementary subgroups that carry arithmetic information.
For a finite-order element x and a prime p, a suitable Sylow p-subgroup of the centraliser
of the p-free part s of x yields a join with ⟨s⟩ that contains x. Every such join is
p-elementary, and when the Sylow subgroup has finite index in the centraliser the join has index
prime to p there. The cyclic subgroups are already elementary, so it is this larger subgroup, not
⟨x⟩, that carries content.
Main definitions #
TauCeti.IsPElementary p G:Gis the internal direct product of a cyclic subgroup of order prime topand ap-subgroup.TauCeti.IsPHyperelementary p G:Ghas a cyclic normal subgroup of order prime topwithp-group quotient.TauCeti.IsElementary G,TauCeti.IsHyperelementary G: the same, for some primep.TauCeti.pElementaryOfSylow s P: for an elementsof order prime topand a Sylowp-subgroupPof the centraliser ofs, the subgroup ofGgenerated bystogether withP.TauCeti.pFreePartHom s P: thep-free part as a homomorphism from that subgroup onto its cyclic factor⟨s⟩, the projection along which class functions of⟨s⟩are pulled back.
Each of the four predicates is restated as an Iff by TauCeti.isPElementary_def,
TauCeti.isPHyperelementary_def, TauCeti.isElementary_def and TauCeti.isHyperelementary_def.
Main results #
TauCeti.isPElementary_of_isCyclic: every finite cyclic group isp-elementary, for every primep; henceTauCeti.isElementary_of_isCyclic, cyclic groups are elementary.TauCeti.IsPElementary.isPHyperelementary:p-elementary groups arep-hyperelementary.TauCeti.IsPElementary.of_injective,TauCeti.IsPHyperelementary.of_injective: both classes are closed under passing to a subgroup, in the form of an injective homomorphism into the group. The subgroup forms areTauCeti.IsPElementary.subgroupandTauCeti.IsPHyperelementary.subgroup.TauCeti.IsHyperelementary.isSolvable: a finite hyperelementary group is solvable, whenceTauCeti.not_isElementary_perm_fin_5, the symmetric group on five letters is not elementary.TauCeti.IsElementary.isNilpotent: a finite elementary group is nilpotent, its cyclic factor being central.TauCeti.isPElementary_sup_of_commute: a cyclic subgroup of order prime topand ap-subgroup that centralises it generate ap-elementary subgroup, andTauCeti.disjoint_of_not_dvd_natCard_of_isPGroupsupplies the disjointness it asks for.TauCeti.isPElementary_pElementaryOfSylowandTauCeti.not_dvd_relIndex_pElementaryOfSylow:TauCeti.pElementaryOfSylow s Pisp-elementary, and, providedPhas finite index in the centraliser ofs, its index there is prime top.TauCeti.exists_mem_pElementaryOfSylow: every elementxlies inTauCeti.pElementaryOfSylowof itsp-free part, for a suitable Sylowp-subgroup.TauCeti.pFreePart_mem_zpowers_of_mem_pElementaryOfSylowandTauCeti.pPart_mem_map_sylow_of_mem_pElementaryOfSylow: thep-free part and thep-part of an element ofTauCeti.pElementaryOfSylow s Plie in its cyclic and its Sylow factor, andTauCeti.pFreePart_mul_of_mem_pElementaryOfSylow: thep-free part is multiplicative there.
Implementation notes #
The p-group condition on the quotient in TauCeti.IsPHyperelementary is spelled out elementwise,
as ∀ g, ∃ k, g ^ p ^ k ∈ C, rather than as IsPGroup p (G ⧸ C). The two are equivalent
(TauCeti.isPHyperelementary_iff_isPGroup_quotient, and the two implications it is assembled from,
TauCeti.isPGroup_quotient_of_forall_exists_pow_mem and
TauCeti.isPHyperelementary_of_isPGroup_quotient), but the quotient G ⧸ C is a group only once
the normality of C is available as an instance, which an existential quantifier inside the
definition cannot supply. The elementwise form keeps the definition free of that bookkeeping and is
also the form the closure proofs use directly.
TauCeti.IsPElementary records the internal direct product decomposition as
Subgroup.IsComplement' together with elementwise commutation, rather than as normality of both
factors: this is the form the proofs consume, and normality of the cyclic factor is derived from it
in TauCeti.normal_of_commute_of_sup_eq_top.
TauCeti.pElementaryOfSylow takes the Sylow subgroup as an argument instead of choosing one: a
centraliser has in general several Sylow p-subgroups, and everything proved about the resulting
subgroup holds for each of them. TauCeti.exists_mem_pElementaryOfSylow makes the choice that puts
a prescribed element inside.
References #
This supplies a prerequisite for
TauCetiRoadmap/RepresentationTheory/InductionRestriction/README.md,
Layer 6 ("Elementary and hyperelementary subgroups"), which asks for exactly these two predicates
"from these orientations, with their basic closure properties", as the indexing sets of Artin's and
Brauer's induction theorems.
- J.-P. Serre, Linear Representations of Finite Groups, Springer GTM 42 (1977), Part II, §10.
A group G is p-elementary when it is the internal direct product of a cyclic subgroup
of order prime to p and a p-subgroup: there are subgroups C and P of G with C cyclic of
order prime to p, P a p-group, every element of C commuting with every element of P, and
C a complement of P.
Instances For
A group G is p-hyperelementary when it has a cyclic normal subgroup C of order prime
to p whose quotient G ⧸ C is a p-group. The condition on the quotient is spelled elementwise,
as ∀ g, ∃ k, g ^ p ^ k ∈ C; see TauCeti.isPGroup_quotient_of_forall_exists_pow_mem for the
identification with IsPGroup p (G ⧸ C).
Instances For
A group is elementary when it is p-elementary for some prime p.
Instances For
A group is hyperelementary when it is p-hyperelementary for some prime p.
Instances For
Restating the definitions #
The four definitions above are public but not @[expose], so downstream modules see their
statements and not their bodies: neither unfold IsPElementary nor simp only [IsPElementary]
is available there. The Iff.rfl lemmas in this section are therefore the only way to eliminate
the predicates outside this file, as Subgroup.isComplement'_def is for Subgroup.IsComplement'
and isPrimePow_def is for IsPrimePow.
TauCeti.IsPElementary p G unfolded: a decomposition of G into a cyclic subgroup of order
prime to p and a complementary p-subgroup commuting with it elementwise.
TauCeti.IsPHyperelementary p G unfolded: a cyclic normal subgroup of order prime to p such
that every element of G has a p-power inside it. See
TauCeti.isPHyperelementary_iff_isPGroup_quotient for the quotient reading of the last clause.
TauCeti.IsElementary G unfolded: p-elementarity for some prime p.
TauCeti.IsHyperelementary G unfolded: p-hyperelementarity for some prime p.
The quotient form of hyperelementarity #
If every element of G has a p-power lying in a normal subgroup C, the quotient G ⧸ C is
a p-group. This is the quotient reading of the last clause of TauCeti.IsPHyperelementary.
G is p-hyperelementary exactly when it has a cyclic normal subgroup C of order prime to
p with IsPGroup p (G ⧸ C). This is the quotient form of the definition, with the normality
witness packaged so that it supplies the group structure on G ⧸ C.
Elementary groups are hyperelementary #
A subgroup centralised elementwise by a complement is normal: the complement form of
TauCeti.normal_of_commute_of_sup_eq_top, matching the shape in which TauCeti.IsPElementary
records its decomposition.
If C and a p-subgroup P are complements centralising each other, every element of G
has a p-power in C: write it as a * b with a ∈ C, b ∈ P, and kill b.
A p-elementary group is p-hyperelementary: the cyclic factor is normal, and the quotient by
it is the p-group factor.
An elementary group is hyperelementary.
Basic examples #
A p-group is p-elementary, with trivial cyclic factor.
A finite cyclic group is p-elementary for every prime p: writing Nat.card G as
p ^ a * m with p ∤ m and taking a generator g, the subgroup generated by g ^ (p ^ a) is
cyclic of order m and the subgroup generated by g ^ m is a p-group of order p ^ a, and the
two are complementary because their orders are coprime.
So the elementary subgroups of a finite group include all its cyclic subgroups: Brauer's induction theorem, indexed by elementary subgroups, refines Artin's, indexed by cyclic ones.
A finite cyclic group is elementary.
A finite cyclic group is hyperelementary.
Closure under subgroups #
p-hyperelementarity passes to subgroups, in the form of an injective homomorphism into the
group.
p-elementarity passes to subgroups, in the form of an injective homomorphism into the group.
The decomposition of a subgroup is inherited factor by factor: the crux is that the subgroup is
still the product of the two intersections, which follows via
TauCeti.exists_zpow_mul_zpow_eq_of_coprime from a Bézout identity between the orders of the two
factors, coprime because one is prime to p and the other is a power of p.
p-hyperelementarity passes to subgroups.
p-elementarity passes to subgroups.
Hyperelementarity passes to subgroups.
Elementarity passes to subgroups.
Solvability #
A finite p-hyperelementary group is solvable: the cyclic normal subgroup C supplied by the
definition is abelian, hence solvable, and the quotient G ⧸ C is a p-group, hence nilpotent and
so solvable too.
A finite hyperelementary group is solvable.
A finite elementary group is solvable.
A finite p-elementary group is nilpotent: the cyclic factor C is central, being
commutative and centralised by the p-group factor, and G ⧸ C is a p-group.
A finite elementary group is nilpotent.
The symmetric group on five letters is not hyperelementary, since it is not solvable. Together
with TauCeti.isElementary_of_isCyclic this pins the two predicates between the cyclic groups and
the solvable ones.
The symmetric group on five letters is not elementary.
p-elementary subgroups built from a Sylow subgroup of a centraliser #
The cyclic subgroups of a finite group are already elementary, so they witness nothing beyond
TauCeti.isPElementary_of_isCyclic. What Brauer's induction theorem needs is a large elementary
subgroup attached to an element: for x : G with p-free part s, some suitable Sylow
p-subgroup of the centraliser of s yields a join with ⟨s⟩ that contains x. Every such join
is p-elementary, and as soon as the Sylow subgroup has finite index in that centraliser the join
has relative index prime to p there, which is the arithmetic the induction theorem runs on.
Two subgroups centralising one another, one cyclic of order prime to p and one a p-group,
generate a p-elementary subgroup.
The p-elementary subgroup attached to an element s of order prime to p together with a
Sylow p-subgroup P of the centraliser of s: the join of the cyclic group generated by s
with P. It is p-elementary by TauCeti.isPElementary_pElementaryOfSylow, and, when P has
finite index in the centraliser, its relative index there is prime to p by
TauCeti.not_dvd_relIndex_pElementaryOfSylow.
Instances For
Every element of ⟨s⟩ commutes with every element centralising s.
The subgroup attached to a p-regular element and a Sylow p-subgroup of its centraliser is
indeed p-elementary.
The subgroup attached to a p-regular element s and a Sylow p-subgroup P of finite index
in its centraliser has index prime to p inside that centraliser: it contains a full Sylow
p-subgroup.
An element of the subgroup attached to s and P factors through the Sylow subgroup.
Every element of TauCeti.pElementaryOfSylow s P is a power of s times an element of the image
of P, and the two factors commute.
The p-part of an element of TauCeti.pElementaryOfSylow s P lies in its Sylow factor.
The factorisation of TauCeti.exists_mul_eq_of_mem_pElementaryOfSylow splits the element into a
factor of order prime to p and a p-element, so it is the factorisation of
TauCeti.eq_pPart.
The p-free part of an element of TauCeti.pElementaryOfSylow s P lies in its cyclic
factor ⟨s⟩. This is the other half of the factorisation
TauCeti.exists_mul_eq_of_mem_pElementaryOfSylow, read through TauCeti.eq_pFreePart.
The p-free part is multiplicative on TauCeti.pElementaryOfSylow s P. The cyclic
factor is central in the subgroup, so the product of two elements factors as the product of their
p-free parts times the product of their p-parts, and that is again a commuting factorisation
of the kind TauCeti.eq_pFreePart recognises.
The p-free part as a homomorphism from TauCeti.pElementaryOfSylow s P onto its cyclic
factor ⟨s⟩. It is the projection of the direct product ⟨s⟩ × P onto its first factor, the one
along which class functions of ⟨s⟩ are pulled back to the p-elementary subgroup.
Instances For
The p-free-part projection is the identity on the cyclic factor.
The p-free-part projection maps onto the cyclic factor.
Membership of an element #
Every element lies in the p-elementary subgroup built from its p-free part.
For every p and every element x, some Sylow p-subgroup of the
centraliser of the p-free part of x produces, by
TauCeti.pElementaryOfSylow, a subgroup containing x. When p is prime and x has finite
order, so that p does not divide the order of TauCeti.pFreePart p x, that subgroup is
p-elementary by TauCeti.isPElementary_pElementaryOfSylow, and when moreover the Sylow subgroup
has finite index in the centraliser, which is not asserted here,
TauCeti.not_dvd_relIndex_pElementaryOfSylow gives the subgroup index prime to p there. For
p not prime or x of infinite order the p-free part is x itself and any Sylow subgroup
works, with no p-elementary conclusion.