Alphabeta Math
Pipeline-generated
How statement and proof provenance work

The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.

  • Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
  • AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
  • AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.

These labels describe origin, not correctness: citations and verification chips remain separate evidence.

✓ 30 results · all verified · 20 also independently AI-judged
Every result on this page is machine-checked by a proof checker and read in full and owner-audited; the judge is an additional, independent cross-model AI review of the proofs. The 10 not AI-judged were verified by owner audit (typically over a confirmed judge false positive), not failures.

Frobenius Groups and the Normal Complement Theorem

1 · Prerequisites

2 · Summary

This page develops the two classical routes to Frobenius' normal p-complement theorem. The first half proves the Frobenius kernel theorem: for a finite group G with a Frobenius complement H, the fixed-point-free set N=(G∖⋃xxHx−1)∪{1} is a normal subgroup and G=N⋊H. The proof is character-theoretic: the adjusted class function φ~=Ind⁡HGθ+φ(1)1G has norm one and is realised by an honest character whose kernel is N. The second half builds the transfer Vφ for a finite-index subgroup, proves independence of the transversal and multiplicativity, derives the cycle-decomposition formula, and uses it for Burnside's normal p-complement theorem, the equivalent forms of having a normal p-complement, the Frobenius normal p-complement theorem in its local and fusion forms, and the automizer criterion.

Conventions fixed here: conjugation is xg=gxg−1, so that (za)b=zba; the p-local quantifier always ranges over the nontrivial normalizers NG(Q) with 1≠Q≤P for a fixed Sylow p-subgroup P; the p′-core Op′(G) and the p-residual Op(G) are the largest normal p′-subgroup and the smallest normal subgroup with p-group quotient. All arguments are choice-free. The local automizer condition gives centralizer conjugacy among Sylow subgroups of each normalizer NG(Q); the intersection-ascent lemma then carries this local conjugacy to fusion control in G. Local suppliers used by the transfer and fusion arguments are proved earlier on the page, before their consumers.

3 · Logical flowchart

4 · Definitions, theorems and proofs

DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Frobenius complement and frobenius group

Definition

Conjugation of subsets. Let G be a group and let S⊆G be a subset. For g∈G write gSg−1:={gsg−1:s∈S}, the image of S under the inner automorphism cg of G. This is the convention of Conjugation x↦gxg−1 is an automorphism and Normal subgroup: invariance under conjugation: conjugation is written on the left, so that gxg−1 is the conjugate of x by g.

Frobenius complement. Let G be a finite group. A subgroup H≤G (Subgroup) with {1}<H<G is a Frobenius complement of G when

H∩gHg−1={1}for every g∈G∖H.

A group that possesses a Frobenius complement is a Frobenius group, and one then says that G is a Frobenius group with complement H. Both subgroups {1} and G are excluded by the hypothesis {1}<H<G, so a Frobenius complement is a nontrivial proper subgroup.

Remarks

  • The condition is symmetric in H and its conjugates. Since the condition is required only for g∉H and is automatic for g∈H (there gHg−1=H, and H∩H=H≠{1} is not required to be trivial), one may equivalently require both NG(H)=H and H∩gHg−1={1} for all g∈G with gHg−1≠H. Replacing g by g−1 shows that H∩gHg−1={1} holds for all g∉H if and only if Hg∩H={1} holds for all g∉H, so the definition is not sensitive to using left rather than right conjugates.

  • H is its own normalizer. If g∈NG(H) (The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup) then gHg−1=H, hence H∩gHg−1=H; since H≠{1} this forces g∈H. Thus for a Frobenius complement the normalizer is as small as possible, NG(H)=H; this is used to count conjugates in Frobenius kernel cardinality.

  • Terminology. The condition is a strong form of malnormality of H; the complement is not assumed to be normal, and the existence of the normal complement of Frobenius kernel theorem is the content of Frobenius' theorem, not part of this definition.

PropositionStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Frobenius permutation action characterization

Statement

Let G be a finite group and let {1}<H<G be a subgroup. Then H is a Frobenius complement of G if and only if the left action of G on the coset space G/H by left multiplication is transitive, is nonregular (that is, not free), and every nonidentity element of G fixes at most one coset.

Facts & Assumptions

Given: A finite group G, a subgroup {1}<H<G, and the rule g⋅xH:=gxH for g,x∈G.

[F1]

A left action of G on a set X is a map G×X→X with e⋅x=x and (gh)⋅x=g⋅(h⋅x) for all g,h∈G, x∈X; the action is transitive when some element of G carries any given point to any other (Left group actions, transitive actions, and faithful actions).

[F2]

For g∈G the left coset is gH={gh:h∈H}, and gH=xH holds exactly when x∈gH (Left and right cosets gH and Hg of a subgroup).

[F3]

The orbit and stabilizer of a point x of a G-set X are G⋅x={g⋅x:g∈G} and Gx={g∈G:g⋅x=x}; the stabilizer is a subgroup (The orbit G⋅x and stabilizer Gx of a point in a group action).

[F4]

An action is free when g⋅x=x implies g=e for all g∈G and all x∈X; equivalently, no nonidentity element fixes any point (A free group action has no nonidentity element fixing a point).

[F5]

For g∈G the fixed-point set is Xg={x∈X:g⋅x=x}; writing "an element fixes at most one coset" means ∣(G/H)g∣≤1 for every g≠1 (The fixed-point sets Xg and XG of a group action).

[F6]

A subgroup {1}<H<G is a Frobenius complement exactly when H∩gHg−1={1} for every g∉H (Frobenius complement and frobenius group).

[F7]

Every orbit of a G-set is in bijection with the left cosets of the stabilizer of a point of it, by gGx↦g⋅x (Orbit-stabiliser: G/Gx→G⋅x, gGx↦g⋅x, is a well-defined bijection).

[F8]

A subgroup H≤G contains e, is closed under products, and is closed under inverses (Subgroup).

[F9]

For a,b∈G one has aH=bH if and only if a−1b∈H, and x∈aH if and only if a−1x∈H (x∈aH iff a−1x∈H, and aH=bH iff a−1b∈H).

Proof

technique · direct
1.1

The rule g⋅xH:=gxH defines a left action of G on G/H: e⋅xH=xH and (gh)⋅xH=ghxH=g⋅(h⋅xH) for all g,h,x∈G, by associativity of the product in G.

F1F2given
2.1

The action of step 1.1 is transitive: given cosets xH and yH, the element g:=yx−1 satisfies g⋅xH=yx−1xH=yH.

F1F2step 1.1algebra
2.2

For k∈G and a coset xH one has k⋅xH=xH if and only if x−1kx∈H: indeed the coset equality kxH=xH is equivalent to (kx)−1x=x−1k−1x∈H by [F9], and x−1k−1x=(x−1kx)−1 is invertible in H exactly when x−1kx∈H, by [F8], so the displayed criterion follows. Consequently the fixed-point set of k is (G/H)k={xH∈G/H:k∈xHx−1}, and k fixes a coset precisely when k lies in the corresponding conjugate xHx−1 of H.

F2F3F5F8F9step 1.1algebra
2.3

The action of step 1.1 is nonregular whenever {1}<H: choose h∈H with h≠1; then h⋅H=H, so some nonidentity element fixes a point of G/H and the action is not free. Conversely, if H={1} then every stabilizer is trivial and the action is free, so nonregularity is exactly the clause {1}<H.

F1F4F8givenchoose
3.1

Suppose H is a Frobenius complement. Let 1≠k∈G and suppose k fixes two distinct cosets xH≠yH. By step 2.2 there are u∈H and v∈H with k=xux−1=yvy−1. Then u=x−1y v (x−1y)−1, so u∈H∩sHs−1 where s:=x−1y. Since xH≠yH we have s=x−1y∉H by [F9], so [F6] gives H∩sHs−1={1} and u=1; hence k=xux−1=1, contradicting k≠1. Therefore every nonidentity element fixes at most one coset.

F6F9step 2.2algebra
3.2

Suppose conversely that every nonidentity element of G fixes at most one coset of H, and let g∉H and z∈H∩gHg−1. By [F6] it suffices to show z=1. The element z lies in H, so z⋅H=H; it also lies in gHg−1, so z⋅gH=gH by step 2.2. Since g∉H the cosets H and gH are distinct by [F9]. Thus the nonidentity element z would fix two distinct cosets, so the hypothesis forces z=1.

F6F2F9step 2.2assume-hyp
4.1

Combining the clauses: if H is a Frobenius complement then the action is transitive (step 2.1), nonregular (step 2.3, as {1}<H), and at most one coset is fixed by each nonidentity element (step 3.1). Conversely, if the action has the three listed properties then the "at most one fixed coset" clause is available and step 3.2 gives H∩gHg−1={1} for every g∉H, that is, H is a Frobenius complement by [F6]. This proves both directions of the stated equivalence. ∎

F6F7step 3.1step 3.2step 2.3
DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Frobenius kernel set

Definition

Let G be a finite group and let H be a Frobenius complement of G (Frobenius complement and frobenius group). The candidate Frobenius kernel set, or simply the kernel set, attached to H is

N:=(G∖ ⁣ ⁣⋃x∈GxHx−1)∪{1}.

Thus an element g∈G lies in N exactly when g=1 or g lies in no conjugate xHx−1 of H. Equivalently, by the fixed-point description of the coset action (Frobenius permutation action characterization), N is the set of elements that either are the identity or fix no coset of H in the left action of G on G/H; the identity is included by hand, because it lies in every conjugate of H.

Two cautions are part of the definition. First, N is defined as a subset of G; no claim that N is a subgroup is built into the notation, and the description "kernel" is provisional. Second, the set is invariant under conjugation: if g∈N and t∈G then tgt−1 lies in no conjugate of H whenever g does, since u(tgt−1)u−1=(ut)g(ut)−1; this setwise invariance is not closure under products.

Remarks

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Frobenius kernel cardinality

Statement

Let G be a finite Frobenius group with complement H and let N=(G∖⋃x∈GxHx−1)∪{1} be its kernel set. Then

∣N∣=[G:H],N∩H={1}.

Facts & Assumptions

Given: A finite group G, a Frobenius complement {1}<H<G, and the kernel set N=(G∖⋃x∈GxHx−1)∪{1}.

[F1]

An element g lies in N exactly when g=1 or g∉xHx−1 for every x∈G (Frobenius kernel set).

[F2]

H∩gHg−1={1} for every g∉H, and {1}<H<G (Frobenius complement and frobenius group).

[F3]

NG(H)={g∈G:gHg−1=H} is a subgroup of G containing H (The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup, CG(x) and NG(H) are subgroups of G).

[F4]

The rule gNG(H)↦gHg−1 is a well-defined bijection G/NG(H)→{gHg−1:g∈G}; for finite G the number of distinct conjugates of H is [G:NG(H)] (The conjugates of H are in bijection with G/NG(H) and, for finite G, number [G:NG(H)]).

[F5]

For a finite group G and H≤G one has ∣G∣=[G:H] ∣H∣ (Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G).

[F6]

[G:H] is the cardinality of the left coset set G/H (The coset set G/H and the index [G:H] of a subgroup).

Proof

technique · direct
1.1

One has NG(H)=H. In one direction H⊆NG(H), since hHh−1=H for h∈H. Conversely let g∈NG(H); then gHg−1=H and hence H=H∩gHg−1. If g∉H then [F2] makes this intersection {1}, contradicting {1}<H; so g∈H.

F2F3given
1.2

Nonidentity elements of distinct conjugates do not overlap: if 1≠y∈xHx−1∩zHz−1 for some x,z∈G, then y=xux−1=zvz−1 for some u,v∈H, whence u=x−1zvz−1x=(x−1z)v(x−1z)−1∈H∩sHs−1 with s=x−1z. Since u≠1, [F2] forces s∈H, that is z∈xH, and then zHz−1=xHx−1.

F2algebra
2.1

Consequently the distinct subgroups of the form xHx−1 are in bijection with the left cosets of H, so there are exactly [G:H] of them: [F4] identifies the set of conjugates with G/NG(H), and step 1.1 together with [F6] identifies the cardinality of that coset space with [G:H].

F3F4F6step 1.1
3.1

Every conjugate xHx−1 has exactly ∣H∣ elements, and by step 1.2 each nonidentity element of the union ⋃x∈GxHx−1 lies in exactly one of the conjugates; the element 1 lies in all of them. Hence the union has 1+[G:H] (∣H∣−1) elements.

step 2.1step 1.2algebra
4.1

Therefore ∣N∣=∣G∣−(1+[G:H](∣H∣−1))+1=∣G∣−[G:H](∣H∣−1), and Lagrange's identity ∣G∣=[G:H] ∣H∣ of [F5] turns this into ∣N∣=[G:H] ∣H∣−[G:H] ∣H∣+[G:H]=[G:H].

F1F5step 3.1algebra
5.1

Finally N∩H={1}: the identity lies in both sets, while a nonidentity element h∈H lies in the conjugate 1H1−1=H, so by the description [F1] of N it is not an element of N. ∎

F1given
DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Induced class functions and restricted class functions

Definition

Let G be a finite group, let H≤G be a subgroup, and let θ∈cf(H) be a complex class function on H (Class functions and the complex vector space cf(G), Subgroup).

Induction. The induced class function Ind⁡HGθ∈cf(G) is defined by the Frobenius formula Ind⁡HGθ(g):=1∣H∣∑x∈Gx−1gx∈Hθ(x−1gx),g∈G. The sum is over the finite set of x∈G with x−1gx∈H, and each summand is a complex number, so the formula is well defined; the normalising factor 1/∣H∣ is the one used for honest induced characters in Frobenius' formula for the character of an induced representation.

Restriction. For ψ∈cf(G) the restricted class function Res⁡HGψ∈cf(H) is the restriction ψ∣H. It is a class function because the conjugating elements in the equation f(hkh−1)=f(k) for k,h∈H are also elements of G.

Elementary properties. The definition is arranged so that the following three statements hold; each is a direct computation from the displayed sum.

  1. Ind⁡HGθ is a class function on G: for t,g∈G the substitution y=tx is a bijection from {x∈G:x−1gx∈H} onto {y∈G:y−1(tgt−1)y∈H}, because (tx)−1(tgt−1)(tx)=x−1gx, and then Ind⁡HGθ(tgt−1)=Ind⁡HGθ(g).
  2. Induction is C-linear in θ: for θ1,θ2∈cf(H) and λ∈C one has Ind⁡HG(θ1+θ2)=Ind⁡HGθ1+Ind⁡HGθ2 and Ind⁡HG(λθ)=λInd⁡HGθ, because both sides are the same finite sum of values of θ1+θ2, respectively λθ.
  3. If θ=χ is the character of a finite-dimensional complex representation of H, then Ind⁡HGχ as defined here is the honest induced character of The induced character Ind⁡HGχ of a complex character: this is exactly the content of the Frobenius formula Frobenius' formula for the character of an induced representation.

The map θ↦Ind⁡HGθ is thus a C-linear map cf(H)→cf(G) extending the honest induction of characters; no claim of positivity, integrality or dependence only on a character is made for a general θ.

Remarks

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Frobenius reciprocity for class functions

Statement

Let G be a finite group, let H≤G be a subgroup, let θ∈cf(H) be a class function on H, and let ψ∈cf(G) be a class function on G. Then

⟨Ind⁡HGθ, ψ⟩G=⟨θ, Res⁡HGψ⟩H.

Facts & Assumptions

Given: A finite group G, a subgroup H≤G, a class function θ on H, a class function ψ on G, and the induced and restricted class functions Ind⁡HGθ and Res⁡HGψ of Induced class functions and restricted class functions.

[F1]

Restriction of class functions is the restriction map ψ↦ψ∣H, induction is the C-linear map θ↦Ind⁡HGθ given by the Frobenius formula, and for an honest character χ of H the class function Ind⁡HGχ is the honest induced character (Induced class functions and restricted class functions).

[F2]

The irreducible complex characters φ1,…,φs of H form an orthonormal basis of cf(H), and the irreducible complex characters χ1,…,χr of G form an orthonormal basis of cf(G) (The irreducible complex characters form an orthonormal basis of cf(G), An irreducible complex character).

[F3]

The inner product ⟨α,β⟩=1∣G∣∑g∈Gα(g)β(g)‾ is linear in its first argument and conjugate-linear in its second, on cf(G) as well as on cf(H) (The standard inner product on cf(G)).

[F4]

For complex characters χ of H and ψ′ of G one has ⟨Ind⁡HGχ,ψ′⟩G=⟨χ,Res⁡HGψ′⟩H (Frobenius reciprocity for complex characters).

[F5]

A class function on a finite group is determined by its values on conjugacy classes, and the space of class functions is a complex vector space (Class functions and the complex vector space cf(G)).

Proof

technique · direct
1.1

Since the irreducible characters form orthonormal bases, there are unique complex numbers a1,…,as and b1,…,br with θ=∑i=1saiφi and ψ=∑j=1rbjχj.

F2F5given
2.1

By the linearity of induction in [F1], Ind⁡HGθ=∑iaiInd⁡HGφi, and the functions Ind⁡HGφi are the honest induced characters of the characters φi; likewise Res⁡HGψ=∑jbjRes⁡HGχj.

F1step 1.1
3.1

The inner product is linear in the first argument and conjugate-linear in the second, by [F3] on G and on H respectively, so bilinearity and the expansions of step 2.1 give ⟨Ind⁡HGθ,ψ⟩G=∑i,jaibj‾⟨Ind⁡HGφi,χj⟩G and ⟨θ,Res⁡HGψ⟩H=∑i,jaibj‾⟨φi,Res⁡HGχj⟩H.

F3step 2.1
3.2

For every pair of indices i,j the honest-character reciprocity [F4] applies to the character φi of H and the character χj of G, giving ⟨Ind⁡HGφi,χj⟩G=⟨φi,Res⁡HGχj⟩H; by step 2.1 the left-hand side is the same as ⟨Ind⁡HGφi,χj⟩G computed with the induced class function.

F1F4step 2.1
4.1

Substituting the identities of step 3.2 into the two expansions of step 3.1 makes the sums equal term by term, so ⟨Ind⁡HGθ,ψ⟩G=⟨θ,Res⁡HGψ⟩H, as claimed. ∎

step 3.1step 3.2
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Zero at identity induction restriction for a frobenius complement

Statement

Let G be a finite Frobenius group with complement H, and let θ∈cf(H) be a complex class function on H with θ(1)=0. Then

Res⁡HGInd⁡HGθ=θ.

Facts & Assumptions

Given: A finite group G, a Frobenius complement {1}<H<G, a class function θ on H with θ(1)=0, and the induced class function Ind⁡HGθ of Induced class functions and restricted class functions.

[F1]

Ind⁡HGθ(h)=1∣H∣∑x∈G: x−1hx∈Hθ(x−1hx) for every h∈H, and Res⁡HG is restriction of functions (Induced class functions and restricted class functions).

[F2]

H∩gHg−1={1} for every g∉H, and {1}<H<G (Frobenius complement and frobenius group).

[F3]

A class function on H satisfies θ(tkt−1)=θ(k) for all t,k∈H (Class functions and the complex vector space cf(G)).

[F4]

H contains the identity, is closed under products, and is closed under inverses (Subgroup).

Proof

technique · direct
1.1

Let h∈H and let x∈G satisfy x−1hx∈H, with h≠1. Then h=x (x−1hx) x−1 lies in H∩xHx−1, and h≠1; so the complement condition forces x∈H. Consequently, for h≠1, the summation index set {x∈G:x−1hx∈H} is contained in H.

F2F4given
1.2

For x∈H one has x−1hx∈H because H is closed under products and inverses, and then θ(x−1hx)=θ(h) because θ is a class function on H.

F3F4given
1.3

For h=1 one has x−11x=1 for every x∈G, so each summand in [F1] is θ(1)=0 and Res⁡HGInd⁡HGθ(1)=0=θ(1).

F1given
2.1

If h∈H∖{1}, the elements x∈G with x−1hx∈H are exactly the elements of H, by steps 1.1 and 1.2; hence Res⁡HGInd⁡HGθ(h)=1∣H∣∑x∈Hθ(h)=θ(h).

F1step 1.1step 1.2algebra
3.1

The two cases h=1 and h≠1 cover every element of H, so the induced class function restricts to θ, as claimed. ∎

step 2.1step 1.3cases
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Frobenius character extension construction

Statement

Let G be a finite Frobenius group with complement H. For a nontrivial irreducible complex character φ of H put θ:=φ−φ(1) 1H and φ~:=Ind⁡HGθ+φ(1) 1G, where 1H and 1G denote the constant function with value 1 on H and on G. Then

φ~∣H=φ,φ~(1)=φ(1),φ~(g)=φ(1)  for every g∈G lying in no conjugate of H.

Facts & Assumptions

Given: A finite group G with Frobenius complement {1}<H<G, a nontrivial irreducible complex character φ of H, and the class functions θ=φ−φ(1)1H and φ~=Ind⁡HGθ+φ(1)1G.

[F1]

Induction Ind⁡HG is C-linear on class functions, is given on g∈G by the Frobenius sum, and agrees with honest induction on honest characters; the constant functions 1H and 1G are the characters of the trivial one-dimensional representations, hence are characters (Induced class functions and restricted class functions, The trivial representation, the regular representation, and permutation representations from finite G-sets).

[F2]

If θ∈cf(H) satisfies θ(1)=0, then Res⁡HGInd⁡HGθ=θ (Zero at identity induction restriction for a frobenius complement).

[F3]

φ is the character of an irreducible complex representation of H, so φ is a class function with φ(1)=dim⁡V≥1; the constant function 1H is the character of the trivial one-dimensional representation, and φ≠1H (An irreducible complex character, For a complex character, χ(1)=dim⁡V, χ is a class function, and ∣χ(g)∣≤χ(1) with equality exactly at scalars, The trivial representation, the regular representation, and permutation representations from finite G-sets).

[F4]

A virtual character of a finite group is an integral linear combination of irreducible complex characters; the class functions on a finite group form a complex vector space (Virtual characters and the character ring R(G) of a finite group, Class functions and the complex vector space cf(G)).

Proof

technique · direct
1.1

The function θ=φ−φ(1)1H is a class function on H with θ(1)=φ(1)−φ(1)⋅1=0; it is a virtual character of H, being the integral combination φ−φ(1)1H of the irreducible character φ and the trivial character 1H.

F3F4givenalgebra
2.1

By the linearity of induction in [F1], Ind⁡HGθ=Ind⁡HGφ−φ(1)Ind⁡HG1H; here Ind⁡HGφ and Ind⁡HG1H are honest characters of G, since φ and 1H are characters of H, so Ind⁡HGθ is a virtual character of G, and so is φ~=Ind⁡HGθ+φ(1)1G.

F1step 1.1
2.2

Since θ(1)=0, [F2] gives Res⁡HGInd⁡HGθ=θ, and therefore φ~∣H=θ+φ(1)1H=φ−φ(1)1H+φ(1)1H=φ.

F2step 1.1algebra
3.1

Evaluating the same identity at the identity gives Ind⁡HGθ(1)=θ(1)=0 and hence φ~(1)=0+φ(1)⋅1=φ(1).

F2step 2.2algebra
4.1

Let g∈G lie in no conjugate of H, that is x−1gx∉H for every x∈G. Then every summand in the Frobenius sum for Ind⁡HGθ(g) is absent, so Ind⁡HGθ(g)=0 and φ~(g)=0+φ(1)⋅1=φ(1). ∎

F1givenalgebra
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-09-27Open item page →

Frobenius character extension is irreducible

Statement

Let G be a finite Frobenius group with complement H and let φ be a nontrivial irreducible complex character of H. Then φ~=Ind⁡HGθ+φ(1)1G with θ=φ−φ(1)1H is an irreducible complex character of G.

Facts & Assumptions

Given: A finite group G with Frobenius complement {1}<H<G, a nontrivial irreducible complex character φ of H, and the class function φ~ of Frobenius character extension construction.

[F1]

Put d=φ(1)∈N>0. The construction gives φ~=Ind⁡HG(φ−d1H)+d1G, φ~∣H=φ and φ~(1)=d, hence Res⁡HGInd⁡HGθ=θ (Frobenius character extension construction). Induction of class functions is linear and sends honest characters to honest characters (Induced class functions and restricted class functions). Thus φ~=Ind⁡HGφ−dInd⁡HG1H+d1G is an integral combination of honest characters. Each honest character has integral coefficients in the irreducible-character basis: its coefficient at χi is its inner product with χi, a dimension of an intertwiner space (The irreducible complex characters form an orthonormal basis of cf(G), The class-function inner product ⟨χV,χW⟩ equals dim⁡Hom⁡G(W,V)). Consequently φ~=∑iniχi with ni∈Z, so it is a virtual character (Virtual characters and the character ring R(G) of a finite group).

[F2]

For class functions α on H and β on G one has ⟨Ind⁡HGα,β⟩G=⟨α,Res⁡HGβ⟩H (Frobenius reciprocity for class functions).

[F3]

The inner product ⟨α,β⟩=1∣G∣∑g∈Gα(g)β(g)‾ is linear in the first argument and conjugate-linear in the second, and ⟨1G,1G⟩=1; the same holds on H (The standard inner product on cf(G)).

[F4]

Irreducible complex characters ψ1,ψ2 of a finite group satisfy ⟨ψ1,ψ2⟩=δ12; the trivial character 1H, being the character of the one-dimensional trivial representation, is irreducible with ⟨1H,1H⟩=1, and since φ≠1H the characters φ and 1H are distinct irreducibles, so ⟨φ,1H⟩=0 and ⟨φ,φ⟩=1 (The first orthogonality relation for irreducible complex characters, An irreducible complex character, The trivial representation, the regular representation, and permutation representations from finite G-sets).

Proof

technique · direct
1.1

With θ=φ−φ(1)1H one computes ⟨θ,θ⟩H=⟨φ,φ⟩−2φ(1)⟨φ,1H⟩+φ(1)2⟨1H,1H⟩=1+0+φ(1)2, using the orthonormality data of [F4] and the linearity of the inner product in its first argument.

F3F4algebra
2.1

Similarly ⟨θ,1H⟩H=⟨φ,1H⟩−φ(1)⟨1H,1H⟩=0−φ(1)=−φ(1), and hence by [F2] ⟨Ind⁡HGθ,1G⟩G=⟨θ,Res⁡HG1G⟩H=−φ(1).

F2F4step 1.1algebra
2.2

By [F2] and the restriction identity of [F1], ⟨Ind⁡HGθ,Ind⁡HGθ⟩G=⟨θ,Res⁡HGInd⁡HGθ⟩H=⟨θ,θ⟩H=1+φ(1)2.

F1F2step 1.1
3.1

Expanding φ~=Ind⁡HGθ+φ(1)1G and using linearity in the first slot and conjugate-linearity in the second (the cross inner products are the equal real number −d) together with ⟨1G,1G⟩=1 gives ⟨φ~,φ~⟩G=⟨Ind⁡HGθ,Ind⁡HGθ⟩G+2φ(1)⟨Ind⁡HGθ,1G⟩G+φ(1)2⟨1G,1G⟩G=(1+φ(1)2)−2φ(1)2+φ(1)2=1.

F3step 2.1step 2.2algebra
4.1

For the expansion φ~=∑iniχi of the virtual character φ~ over the irreducible characters of G in [F1], orthonormality [F4] gives ∑ini2=⟨φ~,φ~⟩G=1; as the coefficients ni are integers, exactly one of them equals ±1 and all others are 0, so φ~=±χ for some irreducible character χ of G.

F1F4step 3.1algebra
5.1

Since φ~(1)=φ(1)≥1 by [F1] while χ(1)≥1 and (−χ)(1)<0, the sign is positive, so φ~=χ is a genuine irreducible character of G. ∎

F1step 4.1given
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Frobenius kernel is an intersection of character kernels

Statement

Let G be a finite Frobenius group with complement H and kernel set N=(G∖⋃x∈GxHx−1)∪{1}. For every nontrivial irreducible complex character φ of H let φ~ be its extension, and put I:={φ∈Irr⁡(H):φ≠1H},M:=⋂φ∈Iker⁡φ~, where ker⁡φ~={g∈G:φ~(g)=φ~(1)}. Then I is nonempty and N=M.

Facts & Assumptions

Given: A finite group G with Frobenius complement {1}<H<G, the kernel set N of Frobenius kernel set, and the family I of nontrivial irreducible complex characters of H with extensions φ~.

[F1]

N consists of 1 and the elements of G that lie in no conjugate xHx−1 of H (Frobenius kernel set).

[F2]

For each φ∈I the class function φ~ satisfies φ~∣H=φ, φ~(1)=φ(1) and φ~(g)=φ(1) for every g∈G lying in no conjugate of H (Frobenius character extension construction).

[F3]

For each φ∈I the class function φ~ is an irreducible complex character of G (Frobenius character extension is irreducible).

[F4]

For a finite-dimensional complex representation ρ with character χ one has ker⁡χ=ker⁡ρ, and ker⁡ρ is a normal subgroup of G; in particular ker⁡φ~ is a normal subgroup of G for each φ∈I (The kernel of a complex character agrees with the kernel of any representation affording it).

[F5]

The intersection of a nonempty family of normal subgroups of G is a normal subgroup of G (The intersection of a nonempty family of normal subgroups is normal).

[F6]

For a finite group H, a subgroup N0≤H is normal if and only if it is an intersection of kernels of irreducible complex characters of H; applying this to N0={1} gives ⋂ψ∈Irr⁡(H)ker⁡ψ={1}, since the intersection over all irreducible characters is contained in any such sub-intersection (The normal subgroups of a finite group are exactly the intersections of kernels of irreducible complex characters).

[F7]

If M′⊴G then xM′x−1=M′ for every x∈G (Normal subgroup: invariance under conjugation).

Proof

technique · direct
1.1

The family I is nonempty: if Irr⁡(H)={1H} were a singleton, then [F6] would give {1}=⋂ψ∈Irr⁡(H)ker⁡ψ=ker⁡1H=H, contradicting {1}<H; hence there is an irreducible character of H different from 1H.

F6given
1.2

N⊆M: let g∈N. If g=1 then φ~(1)=φ~(1) for every φ, so g∈ker⁡φ~ for all φ∈I. If g≠1 then by [F1] the element g lies in no conjugate of H, so [F2] gives φ~(g)=φ(1)=φ~(1) for every φ∈I, that is g∈ker⁡φ~ for every such φ.

F1F2cases
1.3

M∩H={1}: if h∈M∩H then for every φ∈I one has φ(h)=φ~(h)=φ~(1)=φ(1) by [F2], so h∈ker⁡φ, and h∈ker⁡1H holds trivially as well; hence h∈⋂ψ∈Irr⁡(H)ker⁡ψ={1} by [F6].

F2F6algebra
1.4

Every normal subgroup M′ of G with M′∩H={1} satisfies M′⊆N: for x∈G one has M′∩xHx−1=x(x−1M′x∩H)x−1=x(M′∩H)x−1={1} by [F7], so each nonidentity element of M′ lies in no conjugate of H and therefore belongs to N by [F1].

F1F7algebra
2.1

By [F3] each φ~ with φ∈I is an irreducible character, so by [F4] each ker⁡φ~ is a normal subgroup of G; since I is nonempty by step 1.1, [F5] makes M a normal subgroup of G.

F3F4F5step 1.1
3.1

Applying step 1.4 to the normal subgroup M of step 2.1, whose intersection with H is trivial by step 1.3, yields M⊆N; together with step 1.2 this gives N=M, as claimed. ∎

step 1.2step 2.1step 1.3step 1.4
TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Frobenius kernel theorem

Statement

Let G be a finite Frobenius group with complement H and let N=(G∖⋃x∈GxHx−1)∪{1} be the associated kernel set. Then N is a normal subgroup of G.

Facts & Assumptions

Given: A finite group G with Frobenius complement {1}<H<G and the kernel set N with its associated family I of nontrivial irreducible characters of H.

[F1]

N=⋂φ∈Iker⁡φ~ for a nonempty family {φ~:φ∈I} of irreducible complex characters of G (Frobenius kernel is an intersection of character kernels).

[F2]

If ρ is a finite-dimensional complex representation of G with character χ, then ker⁡χ=ker⁡ρ⊴G (The kernel of a complex character agrees with the kernel of any representation affording it).

[F3]

The intersection of a nonempty family of normal subgroups of G is again a normal subgroup of G (The intersection of a nonempty family of normal subgroups is normal).

Proof

technique · direct
1.1

By [F1], N is the intersection over the nonempty family of kernels ker⁡φ~ of irreducible characters φ~ of G.

F1given
2.1

Each ker⁡φ~ occurring in [F1] is a normal subgroup of G: it is the kernel of the representation ρ affording the character φ~, hence equals ker⁡ρ, which is normal by [F2].

F2step 1.1
3.1

Therefore N is an intersection of a nonempty family of normal subgroups of G, so by [F3] it is a subgroup of G and is normal in G. In particular the kernel set is closed under products and inverses, a fact that the counting argument of Frobenius kernel cardinality could not supply. ∎

F1F3step 1.1step 2.1
CorollaryStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Frobenius semidirect product decomposition

Statement

Let G be a finite Frobenius group with complement H and kernel set N=(G∖⋃x∈GxHx−1)∪{1}. Then G is the internal semidirect product G=N⋊H, that is N⊴G, G=NH and N∩H={1}; moreover ∣N∣=[G:H], and N is the unique normal subgroup M⊴G with MH=G and M∩H={1}.

Facts & Assumptions

Given: A finite group G with Frobenius complement {1}<H<G, and the kernel set N of Frobenius kernel set.

[F1]
[F2]

∣N∣=[G:H] and N∩H={1} (Frobenius kernel cardinality).

[F3]

N consists of 1 and the elements lying in no conjugate of H (Frobenius kernel set).

[F4]

If H≤G and N⊴G then HN is a subgroup of G (If H≤G and N⊴G, then HN is a subgroup and H∩N⊴H).

[F5]

If N⊴G and H≤G then H/(H∩N)≅HN/N, hence ∣HN∣ ∣H∩N∣=∣H∣ ∣N∣ (Second isomorphism theorem for groups: H/(H∩N)≅HN/N, Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G).

[F7]

G is the internal semidirect product of N by H exactly when N⊴G, G=NH and N∩H={1} (An internal semidirect product and a complement to a normal subgroup).

[F8]

If M⊴G then xMx−1=M for every x∈G, so M∩xHx−1=x(x−1Mx∩H)x−1=x(M∩H)x−1 (Normal subgroup: invariance under conjugation).

[F9]

If M⊴G, MH=G and M∩H={1} then every g∈G has a unique expression g=mh with m∈M, h∈H: from mh=m′h′ one gets m′−1m=h′h−1∈M∩H={1} (Subgroup).

Proof

technique · direct
1.1

NH is a subgroup of G, since N⊴G and H≤G; its order satisfies ∣NH∣ ∣N∩H∣=∣N∣ ∣H∣ by the second isomorphism theorem together with Lagrange.

F1F4F5
1.2

For uniqueness, let M⊴G satisfy MH=G and M∩H={1}. By [F8], M∩xHx−1=x(M∩H)x−1={1} for every x∈G, so no nonidentity element of M lies in a conjugate of H; hence M⊆N by the description [F3] of N.

F3F8assume-hyp
2.1

Since N∩H={1}, step 1.1 gives ∣NH∣=∣N∣ ∣H∣=[G:H] ∣H∣=∣G∣ by [F2] and [F6]; as NH⊆G is a subgroup with as many elements as G, it equals G.

F2F6step 1.1algebra
3.1

Together with N⊴G of [F1] and N∩H={1} of [F2], step 2.1 exhibits G as the internal semidirect product N⋊H in the sense of [F7], and ∣N∣=[G:H] is [F2].

F1F2F7step 2.1
4.1

Since M∩H={1} and MH=G, the uniqueness of the expression g=mh of [F9] applies with M in place of N and gives ∣G∣=∣M∣ ∣H∣, so ∣M∣=∣G∣/∣H∣=[G:H]=∣N∣ by [F2] and [F6]; with M⊆N from step 1.2 this forces M=N. ∎

F2F6F9step 1.2algebra
PropositionStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Frobenius groups and fixed point free actions

Statement

Let N and H be subgroups of a finite group G with G=N⋊H, 1<N and 1<H, and let H act on N by conjugation, h⋅n:=hnh−1.

  1. If G is a Frobenius group with complement H (so that N is its Frobenius kernel), then every 1≠h∈H fixes only the identity of N: the conjugation action of H on N∖{1} is free.
  2. Conversely, if the conjugation action of H on N∖{1} is free, then H is a Frobenius complement of G.

Facts & Assumptions

Given: A finite group G with subgroups N,H such that N⊴G, G=NH and N∩H={1}, with 1<N and 1<H, and the conjugation action h⋅n=hnh−1 of H on N.

[F1]

N⊴G, G=NH, N∩H={1}, and H is called a complement to N; these are exactly the internal-semidirect-product conditions (An internal semidirect product and a complement to a normal subgroup).

[F2]

N⊴G means gNg−1=N for every g∈G (Normal subgroup: invariance under conjugation).

[F3]

A subgroup {1}<H<G is a Frobenius complement exactly when H∩gHg−1={1} for every g∉H (Frobenius complement and frobenius group).

[F4]

For a finite Frobenius group with complement H the kernel set N is normal and G=N⋊H with N∩H={1} and ∣N∣=[G:H] (Frobenius semidirect product decomposition, Frobenius kernel theorem).

[F5]

H and N are subgroups: each contains the identity, is closed under products, and is closed under inverses (Subgroup).

Proof

technique · direct
1.1

Suppose first that G is a Frobenius group with complement H, so that N is its Frobenius kernel, N⊴G and N∩H={1} by [F4]. Let 1≠h∈H and 1≠n∈N satisfy h⋅n=n, that is hnh−1=n. Then hn=nh and therefore h=n−1hn∈H∩n−1Hn.

F2F4givenalgebra
1.2

Suppose conversely that the conjugation action of H on N∖{1} is free. Let g∈G∖H. By [F1] write g=nh with n∈N, h∈H; if n=1 then g=h∈H, so n≠1. Conjugating, gHg−1=nhHh−1n−1=nHn−1, because hHh−1=H: thus H∩gHg−1=H∩nHn−1.

F1F5assume-hypalgebra
1.3

Let 1≠n∈N and suppose 1≠x∈H∩nHn−1. Then y:=n−1xn satisfies y∈H and x=nyn−1, so xy−1=xn−1x−1 n. Here xn−1x−1∈N by [F2] and n∈N, so xy−1∈N; as also xy−1∈H, the triviality of N∩H forces xy−1=1, that is y=x. Thus n−1xn=x, i.e. xn=nx and x⋅n=n: the nonidentity element x∈H fixes the nonidentity element n∈N.

F2F5assume-hypalgebra
2.1

In the situation of step 1.1 the element n satisfies n∉H: otherwise n∈N∩H={1}, contrary to n≠1; hence also n−1∉H. The complement condition [F3] therefore gives H∩n−1Hn={1}, so step 1.1 forces h=1, contradicting h≠1. Hence no nonidentity n∈N is fixed by a nonidentity h∈H, which is claim 1.

F3F4step 1.1contradiction
2.2

Step 1.3 contradicts freeness of the action on N∖{1}; therefore H∩nHn−1={1} for every 1≠n∈N. By step 1.2 every g∉H has H∩gHg−1=H∩nHn−1 for some n≠1 in N, so H∩gHg−1={1} for every g∉H.

step 1.2step 1.3contradiction
3.1

Finally 1<H holds by hypothesis and H≠G: if H=G then N=N∩G=N∩H={1}, contradicting 1<N. Hence {1}<H<G and H∩gHg−1={1} for all g∉H, so H is a Frobenius complement of G by [F3], which is claim 2. ∎

F1F3step 2.2given
DefinitionDefinition: Literature-sourcedProof: Not applicableaudited 2026-09-27Open item page →

The p-prime core of a finite group

Definition

Let G be a finite group (Group and abelian group, The cardinality ∣A∣ of a finite set) and let p be a prime (Prime and composite integers: p is prime when p>1 and its only positive divisors are 1 and p).

p-element terminology. An element g∈G of finite order is a p-element when ord⁡(g) is a power of p, and a p′-element when p∤ord⁡(g) (The order ∣G∣ of a finite group and the order ord⁡(g) of an element, with ord⁡(g)=∞ when no positive power of g is the identity, A finite p-group has order pn for a prime p and some n∈N). A subgroup K≤G is a p-subgroup when ∣K∣ is a power of p, and a p′-subgroup when p∤∣K∣; the trivial subgroup is both. The terminology is used without further comment throughout the normal-complement material of this page.

The p′-core. Call a subgroup N≤G of G a normal p′-subgroup when N⊴G and p∤∣N∣ (Normal subgroup: invariance under conjugation). The p′-core of G is

Op′(G):=⟨ ⋃{N≤G:N⊴G, p∤∣N∣} ⟩,

the subgroup generated by all normal p′-subgroups of G (The subgroup ⟨S⟩ generated by a subset, the cyclic subgroup ⟨g⟩, and cyclic groups). It is the unique largest normal p′-subgroup of G: it is normal in G, its order is prime to p, and every normal p′-subgroup of G is contained in it.

Why the definition is well posed. The family S={N≤G:N⊴G, p∤∣N∣} is nonempty ({1}∈S, since p∤1) and finite: every member is a subset of the finite set G, and G has finitely many subsets (The cardinality ∣A∣ of a finite set, ∣P(A)∣=2∣A∣ for finite A). Write S={N1,…,Nr}.

Products stay in the family. If M,N∈S, then MN is a subgroup of G (If H≤G and N⊴G, then HN is a subgroup and H∩N⊴H), it is normal because g(MN)g−1=gMg−1 gNg−1=MN for every g∈G (Normal subgroup: invariance under conjugation, Conjugation x↦gxg−1 is an automorphism), and its order divides ∣M∣ ∣N∣: by the second isomorphism theorem M/(M∩N)≅MN/N, so ∣MN∣=∣M∣ ∣N∣/∣M∩N∣ (Second isomorphism theorem for groups: H/(H∩N)≅HN/N, Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G). A divisor of the p′-number ∣M∣ ∣N∣ is again prime to p: if the prime p divided ∣MN∣ it would divide ∣M∣ ∣N∣ and hence, by Euclid's lemma, one of ∣M∣, ∣N∣ (Euclid's lemma: if p is prime and p∣ab then p∣a or p∣b, Divisibility is reflexive and transitive on Z, and is linear: if d∣a and d∣b then d∣ax+by for all integers x,y; also d∣a implies d∣ac, −d∣a and d∣−a). So MN∈S.

The generated subgroup is a member. By the product closure just proved, T:=N1⋯Nr belongs to S, by finite induction. This subgroup contains each Ni (insert identities in all other factors), hence contains ⟨⋃S⟩ by the defining minimality of the generated subgroup (The subgroup ⟨S⟩ generated by a subset, the cyclic subgroup ⟨g⟩, and cyclic groups). Conversely every factor Ni lies in that generated subgroup, so their product T does too. Thus Op′(G)=T∈S; in particular Op′(G)=⋃S, because T is itself a member of the family and contains every member.

Largest and unique. As a member of S, Op′(G) is a normal p′-subgroup of G, and it contains every N∈S by construction; a normal p′-subgroup is by definition a member of S. Hence Op′(G) is the largest normal p′-subgroup, and it is the only one with that property, since two normal p′-subgroups each contain the other.

DefinitionDefinition: Literature-sourcedProof: Not applicableaudited 2026-09-27Open item page →

Normal p complement and p nilpotent group

Definition

Let G be a finite group and let p be a prime.

p-prime terminology. The p-element and p′-element language, for elements and for subgroups, is fixed once and for all in The p-prime core of a finite group and is used without further comment throughout the normal-complement material of this page.

Normal p-complement. A normal p-complement of G is a normal subgroup K⊴G (Normal subgroup: invariance under conjugation) such that

p∤∣K∣and[G:K] is a power of p.

Thus K is a normal p′-subgroup whose index is a p-power. A group possessing a normal p-complement is called p-nilpotent.

Semidirect form. Let P∈Syl⁡p(G) (Sylow p-subgroups of a finite group). If K is a normal p-complement then [G:K]=∣P∣ by Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G and Sylow p-subgroups of a finite group, so K∩P={1} and KP=G with K⊴G; that is, G=K⋊P (An internal semidirect product and a complement to a normal subgroup). Conversely, if G=K⋊P for some P∈Syl⁡p(G), then K is a normal p′-subgroup and [G:K]=∣P∣ is a p-power, so K is a normal p-complement. In particular the definition is equivalent to the existence of a semidirect decomposition of G with normal factor a p′-subgroup.

Remarks

  • Boundary cases. Both extremes are included. If p∤∣G∣ then P={1} and K=G is a normal p-complement, so such a group is p-nilpotent vacuously; if G is a p-group then P=G and K={1} is a normal p-complement. In neither case is the complement required to be proper or nontrivial, in contrast to the Frobenius complement of Frobenius complement and frobenius group.

  • Uniqueness. A normal p-complement, when it exists, is unique: if K1 and K2 are both of p′-order with p-power index, then K1K2/K2≅K1/(K1∩K2) is both a subgroup of the p-group G/K2 and a quotient of the p′-group K1. It is therefore trivial; hence K1⊆K2 and equality follows from ∣K1∣=∣K2∣=∣G∣/∣P∣. The identified complement is the p′-core Op′(G) of The p-prime core of a finite group, and the characterisations of Equivalent forms of having a normal p complement record further equivalent forms.

  • Relation to the Frobenius condition. A finite Frobenius group with kernel N and complement H has N as a normal p-complement whenever p∤∣N∣ and ∣H∣ is a power of p: then [G:N]=∣H∣ is a p-power. In particular, Frobenius groups with a p′-kernel and a p-group complement are p-nilpotent in this sense; the link is drawn in Frobenius normal two complement for S_3 ↗.

DefinitionDefinition: Literature-sourcedProof: Not applicableaudited 2026-09-27Open item page →

Transfer homomorphism for a finite index subgroup

Definition

Let G be a finite group, let H≤G be a subgroup, let A be an abelian group written multiplicatively, and let φ:H→A be a group homomorphism (Monoid homomorphism and group homomorphism). Write H\G for the set of right cosets Ht of H in G (Left and right cosets gH and Hg of a subgroup), and for each coset α=Htα∈H\G choose a representative tα.

The transfer. For x∈G the transfer of φ is

Vφ(x):=∏α∈H\Gφ ⁣(tα x tαx−1),where αx:=Htαx.

Two comments make the displayed formula a definition rather than a shorthand.

  1. Each factor lies in H, so φ is evaluated inside its domain. The coset αx=Htαx is a right coset of H, with the chosen representative tαx; hence tαx tαx−1∈H, because tαx and tαx represent the same right coset (x∈aH iff a−1x∈H, and aH=bH iff a−1b∈H).
  2. The product is well defined although the coset set is unordered. The index set H\G is finite, and A is abelian, so the product of the finitely many elements φ(tαxtαx−1)∈A does not depend on the order in which the factors are written; the factor attached to the coset α is determined by α, x, the chosen representatives and φ.

The definition of Vφ therefore depends on the chosen transversal {tα}; the fact that it does not depend on that choice is proved in Transfer is independent of the transversal, and the fact that Vφ:G→A is a homomorphism is proved in Transfer is a homomorphism.

Remarks

  • Right action convention. The symbol αx denotes right multiplication on the coset, αx=Htα⋅x, and the coset assignment α↦αx is a permutation of the finite set H\G. The transfer is thus built from the right action of G on its right cosets of H, the choice that makes every factor tαxtαx−1 lie in H; with left cosets the analogous factor is txα−1xtα, where α=tαH and xα=xtαH=txαH.

  • Abelian target. Abelianness of A is used only to make the ordering of the product irrelevant; the elements φ(⋅) need not commute in a nonabelian target, and the construction is not made there.

  • Group-theoretic role. The transfer is the tool that converts information about the p-part of G into a homomorphism into an abelian quotient of a Sylow subgroup; it is applied in Burnside normal p complement theorem and Fusion control forces trivial Sylow intersection with the p residual.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Transfer is independent of the transversal

Statement

Let G be a finite group, let H≤G be a subgroup, let A be an abelian group written multiplicatively, and let φ:H→A be a homomorphism (Transfer homomorphism for a finite index subgroup). Let {tα}α∈H\G and {tα′}α∈H\G be two choices of representatives of the right cosets of H in G, and let

V(x):=∏α∈H\Gφ(tα x tαx−1),V′(x):=∏α∈H\Gφ(tα′ x tαx′−1)

be the two products formed from them, where αx:=Htαx. Then V(x)=V′(x) for every x∈G. In particular the transfer Vφ is a well-defined function G→A depending only on φ.

Facts & Assumptions

Given: A finite group G, a subgroup H≤G, an abelian group A, a homomorphism φ:H→A, and two transversals {tα}, {tα′} of the right cosets α∈H\G as in Transfer homomorphism for a finite index subgroup.

[F1]

For every α the coset αx=Htαx is a right coset of H, the assignment α↦αx is a permutation of H\G, each tαxtαx−1 lies in H, and the finite product ∏α∈H\Gaα of elements of the abelian group A is independent of the order of its factors (Transfer homomorphism for a finite index subgroup).

[F2]

If u,v∈G represent the same right coset of H, that is Hu=Hv, then uv−1∈H; conversely uv−1∈H implies Hu=Hv (x∈aH iff a−1x∈H, and aH=bH iff a−1b∈H, Left and right cosets gH and Hg of a subgroup).

[F4]

The map α↦αx is a bijection of the finite set H\G, so a product indexed by H\G may be reindexed along it (The coset set G/H and the index [G:H] of a subgroup, Transfer homomorphism for a finite index subgroup).

Proof

technique · direct
1.1

For every α∈H\G one has tα′=hαtα for some hα∈H: both tα′ and tα represent the coset α, so tα′tα−1∈H by [F2], and hα:=tα′tα−1 is the desired element.

F2given
2.1

For x∈G and α∈H\G the product tα′xtαx′−1 equals hα tαx tαx−1 hαx−1, because tα′=hαtα and tαx′=hαxtαx by step 1.1; hence φ(tα′xtαx′−1)=φ(hα) φ(tαxtαx−1) φ(hαx)−1 by [F3].

F3step 1.1algebra
2.2

The reindexing α↦αx is a bijection of H\G by [F4], so ∏αφ(hαx)−1=∏βφ(hβ)−1=(∏βφ(hβ))−1 by [F3].

F3F4step 1.1
3.1

Consequently V′(x)=(∏αφ(hα))(∏αφ(tαxtαx−1))(∏αφ(hαx)−1): the product over α of the three factors of step 2.1 may be rearranged because A is abelian, by [F1].

F1step 2.1
4.1

Therefore V′(x)=(∏αφ(hα)) V(x) (∏αφ(hα))−1=V(x), the two outer factors cancelling because multiplication in the abelian group A commutes.

F1step 3.1step 2.2algebra
5.1

Since x∈G was arbitrary, V′=V; the transfer is therefore independent of the choice of transversal. ∎

step 4.1given
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Transfer is a homomorphism

Statement

Let G be a finite group, H≤G, A an abelian group written multiplicatively, φ:H→A a homomorphism, and Vφ:G→A the transfer of Transfer homomorphism for a finite index subgroup, formed with a transversal {tα} of the right cosets H\G (the result is independent of the transversal by Transfer is independent of the transversal). Then

Vφ(xy)=Vφ(x) Vφ(y)for all x,y∈G;

that is, the transfer is a group homomorphism G→A.

Facts & Assumptions

Given: A finite group G, a subgroup H≤G, an abelian group A, a homomorphism φ:H→A, a transversal {tα}α∈H\G and the transfer V(x)=∏α∈H\Gφ(tαxtαx−1) of Transfer homomorphism for a finite index subgroup.

[F1]

The assignment α↦αx=Htαx is a right action of G on the finite set H\G, so α(xy)=(αx)y and α↦αx is a permutation of H\G with inverse α↦αx−1; furthermore tαxtαx−1∈H (Transfer homomorphism for a finite index subgroup).

[F2]

The transfer does not depend on the transversal (Transfer is independent of the transversal).

[F4]

The product of finitely many elements of the abelian group A is independent of the order of the factors, and H\G is finite (Transfer homomorphism for a finite index subgroup, Left and right cosets gH and Hg of a subgroup).

Proof

technique · direct
1.1

For x,y∈G and α∈H\G, the identity tαxy tαxy−1=(tαxtαx−1)(tαxytαxy−1) holds: the middle factor tαx−1tαx cancels, and (αx)y=α(xy) by [F1].

F1algebra
2.1

Both bracketed factors of step 1.1 lie in H by [F1], so applying φ gives φ(tαxy tαxy−1)=φ(tαxtαx−1) φ(tαxytαxy−1) by [F3].

F3step 1.1
3.1

Hence V(xy)=∏αφ(tαxtαx−1) φ(tαxytαxy−1), the product of the two factors over α; since A is abelian this equals (∏αφ(tαxtαx−1))(∏αφ(tαxytαxy−1)) by [F4].

F4step 2.1
4.1

The reindexing β:=αx runs over H\G as α does, by the permutation property in [F1], so ∏αφ(tαxytαxy−1)=∏βφ(tβytβy−1)=V(y).

F1step 3.1
5.1

Since V does not depend on the chosen transversal by [F2], the value V(x) is well defined for every x∈G; combining steps 3.1 and 4.1, V(xy)=V(x) V(y) for all x,y∈G, so V is a homomorphism. ∎

F2step 3.1step 4.1
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Transfer cycle decomposition formula

Statement

Let G be a finite group, H≤G, A an abelian group written multiplicatively, φ:H→A a homomorphism, and Vφ:G→A the transfer of Transfer homomorphism for a finite index subgroup, which is independent of the transversal by Transfer is independent of the transversal. Let x∈G and let

H\G=C1⊔⋯⊔Cr

be the decomposition of the finite set of right cosets into the orbits of the right multiplication action of the cyclic subgroup ⟨x⟩≤G (The subgroup ⟨S⟩ generated by a subset, the cyclic subgroup ⟨g⟩, and cyclic groups, ⟨g⟩={ gn:n∈Z }, and every cyclic group is abelian). Put ni:=∣Ci∣ and choose ti∈G with Hti∈Ci. Then

Vφ(x)=∏i=1rφ(ti xni ti−1).

Facts & Assumptions

Given: A finite group G, a subgroup H≤G, an abelian group A, a homomorphism φ:H→A, an element x∈G, and the transfer V=Vφ of Transfer homomorphism for a finite index subgroup.

[F1]

For α∈H\G the coset αt=Htαt is defined for every t∈G, the assignment α↦αt is a right action of G on the finite set H\G, each factor tαstαs−1 (s∈G) lies in H, and products of finitely many elements of the abelian group A are independent of the order of the factors (Transfer homomorphism for a finite index subgroup).

[F2]

The transfer V(x)=∏α∈H\Gφ(tαxtαx−1) does not depend on the transversal (Transfer is independent of the transversal).

[F3]

If u,v∈G satisfy Hu=Hv then uv−1∈H, and conversely (x∈aH iff a−1x∈H, and aH=bH iff a−1b∈H).

[F7]

Every nonempty set of positive integers has a least element, and for integers k and n≥1 there are q,r∈Z with k=qn+r and 0≤r<n (The well-ordering principle, Division with remainder in Z: for a∈Z and b>0 there are unique q,r∈Z with a=qb+r and 0≤r<b).

Proof

technique · direct
1.1

Fix α∈H\G. Since H\G is finite, the elements α,αx,αx2,… cannot all be distinct, so αxi=αxj for some 0≤i<j; applying the inverse permutation α↦αx−i (which is the action of x−i) gives αxj−i=α, an equality of the form αxn=α with n=j−i≥1. By [F7] there is a least positive integer n(α) with αxn(α)=α.

F1F5F7algebra
2.1

For every k∈Z one has αxk=α if and only if n(α)∣k: if k=qn(α) then αxk=α by [F5] and step 1.1 (including q<0, since αx−n(α)=α as well), while for arbitrary k writing k=qn(α)+r with 0≤r<n(α) by [F7] gives αxr=α, so minimality forces r=0.

F5F7step 1.1
3.1

The orbit of α under ⟨x⟩ is C(α)={αxk:k∈Z}, and by step 2.1 it equals {α,αx,…,αxn(α)−1}, whose n(α) elements are pairwise distinct. In particular αxn(α)=α and ∣C(α)∣=n(α).

step 2.1
4.1

The orbit decomposition H\G=C1⊔⋯⊔Cr of [F6] is therefore a decomposition into finitely many sets Ci each of the form Ci={αi,αix,…,αixni−1} with ni=∣Ci∣ and αixni=αi; write αi=Hti.

F6step 3.1given
5.1

The rule tαixj:=tixj for 0≤j<ni defines a transversal of H\G: indeed every coset of H\G lies in exactly one Ci and hence equals exactly one αixj, and tixj represents it, because Htixj=αixj by [F1].

F1step 4.1
6.1

Because V is independent of the transversal by [F2], it may be computed with the transversal of step 5.1: V(x)=∏i=1r∏j=0ni−1φ(tαixj x t(αixj)x−1), where α:=αixj and αx=αixj+1.

F1F2step 5.1
7.1

For 0≤j<ni−1 one has t(αixj)x=tαixj+1=tixj+1=tixjx=tαixjx, so the corresponding factor equals φ(tαixjx(tαixjx)−1)=φ(e)=1 by [F4].

F4step 6.1algebra
7.2

For j=ni−1 one has αixj+1=αixni=αi by step 4.1, so t(αixni−1)x=tαi=ti and the corresponding factor equals φ(tixni−1xti−1)=φ(tixniti−1), an element of H because Htixni=Hti by step 4.1 and [F3].

F3step 4.1step 6.1algebra
8.1

Multiplying the contributions of steps 7.1 and 7.2 over all i and j, all factors with j<ni−1 are 1 and the remaining one for each i is φ(tixniti−1); hence V(x)=∏i=1rφ(tixniti−1). ∎

step 7.1step 7.2given
PropositionStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-09-27Open item page →

Equivalent forms of having a normal p complement

Statement

Let G be a finite group and let p be a prime. The following are equivalent.

(a) G has a normal p-complement. (b) G has a normal p′-subgroup K with [G:K] a power of p. (c) There is a Sylow p-subgroup Q of G and an epimorphism θ:G→Q. (d) The product of any two p′-elements of G is a p′-element. (e) Every p′-element of G lies in Op′(G).

Moreover, if these conditions hold, then Op′(G) is the normal p-complement of G, it is the unique normal p-complement of G, it equals the set of p′-elements of G and the subgroup generated by them, and it is the kernel of every epimorphism θ as in (c).

Facts & Assumptions

Given: A finite group G and a prime p, with p′-element terminology and the p′-core as in The p-prime core of a finite group and normal p-complement as in Normal p complement and p nilpotent group.

[F1]

A normal p-complement is a normal subgroup K⊴G with p∤∣K∣ and [G:K] a power of p; for P∈Syl⁡p(G) it is equivalent to G=KP with K∩P={1}, and [G:K]=∣P∣ (Normal p complement and p nilpotent group, An internal semidirect product and a complement to a normal subgroup, Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G).

[F2]

Op′(G) is a normal p′-subgroup of G containing every normal p′-subgroup of G (The p-prime core of a finite group).

[F3]

G has a Sylow p-subgroup P, of order pa where ∣G∣=pam with p∤m; thus ∣G∣=[G:P] ∣P∣ and p∤[G:P] (Sylow I: every finite group has a Sylow p-subgroup, Sylow p-subgroups of a finite group, Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G).

[F4]

For K⊴G the quotient G/K is a group with ∣G/K∣=[G:K], and the natural map G→G/K is an epimorphism with kernel K (The quotient group G/N and coset product (gN)(hN)=ghN, If [G:N] is finite then ∣G/N∣=[G:N]; for finite G this equals ∣G∣/∣N∣).

[F5]

For a homomorphism θ:G→H one has G/ker⁡θ≅im⁡θ, ker⁡θ⊴G and im⁡θ≤H; hence ∣G∣=∣ker⁡θ∣⋅∣im⁡θ∣ (First isomorphism theorem for groups: G/ker⁡f≅im⁡f, The image of a group homomorphism is a subgroup and its kernel is a normal subgroup, If [G:N] is finite then ∣G/N∣=[G:N]; for finite G this equals ∣G∣/∣N∣).

[F6]

If K⊴G and P≤G then P/(P∩K)≅PK/K and PK≤G, so ∣PK∣ ∣P∩K∣=∣P∣ ∣K∣ (Second isomorphism theorem for groups: H/(H∩N)≅HN/N, If H≤G and N⊴G, then HN is a subgroup and H∩N⊴H).

[F9]

If the prime p divides the order of a finite group H, then H has an element of order p (Cauchy's theorem: if a prime p divides ∣G∣, then G has an element of order p).

[F10]

The subgroup generated by a set Y consists of finite products of elements of Y and their inverses, and conjugation is an automorphism of G (The subgroup ⟨S⟩ generated by a subset, the cyclic subgroup ⟨g⟩, and cyclic groups, Conjugation x↦gxg−1 is an automorphism).

Proof

technique · direct
1.1

(a) implies (b) directly: a normal p-complement is by definition a normal p′-subgroup of p-power index.

F1given
1.2

(b) implies (c): let K be a normal p′-subgroup with [G:K]=pr, so that G/K is a finite group of order pr by [F4]. By [F6] applied to K⊴G and a Sylow P of G given by [F3], P/(P∩K)≅PK/K≤G/K; here ∣P∩K∣ divides ∣P∣=pa and also divides ∣K∣, which is prime to p, so ∣P∩K∣=1 by [F8]. Hence ∣PK/K∣=∣P∣=pa divides ∣G/K∣=pr, so r≥a. Since ∣K∣ divides ∣G∣=pam and is prime to p, every prime divisor of ∣K∣ divides m, so ∣K∣ divides m and m/∣K∣ is an integer prime to p; the identity pr=∣G∣/∣K∣=pa(m/∣K∣) therefore forces r=a and ∣K∣=m. Hence ∣PK∣=∣P∣ ∣K∣/∣P∩K∣=pa∣K∣=pam=∣G∣ by [F6], so PK=G.

F3F4F6F8algebra
1.3

(d) implies (e): let x be a p′-element and let Y be the set of conjugates of x; by (d) and [F11] every finite product of elements of Y is a p′-element, so every element of ⟨Y⟩ is a p′-element by [F10]. If a prime ℓ divided ∣⟨Y⟩∣, then ⟨Y⟩ would contain an element of order ℓ by [F9], which forces ℓ≠p; hence p∤∣⟨Y⟩∣, and ⟨Y⟩ is a p′-subgroup. It is normal: g⟨Y⟩g−1=⟨gYg−1⟩=⟨Y⟩ for every g∈G, by [F10] and the fact that Y consists of all conjugates of x. Hence ⟨Y⟩ is a normal p′-subgroup and ⟨Y⟩≤Op′(G) by [F2], so x∈Op′(G).

F2F9F10F11assume-hyp
1.4

(e) implies (a): write ∣G∣=pam with p∤m as in [F3], and let Q be a Sylow q-subgroup of G for each prime q≠p dividing ∣G∣, which exists by [F3] applied to Q in place of P (Sylow I: every finite group has a Sylow p-subgroup). Every element of Q has order dividing ∣Q∣=qb, hence prime to p, so every element of Q lies in Op′(G) by (e); thus Q≤Op′(G) and ∣Q∣ divides ∣Op′(G)∣ by [F8]. Multiplying over the primes q≠p, the p′-number m divides ∣Op′(G)∣; since Op′(G) is a p′-group by [F2], ∣Op′(G)∣ divides m, so ∣Op′(G)∣=m.

F2F3F8assume-hyp
1.5

A subgroup is a p′-group exactly when all its elements are p′-elements; hence, by [F2] and (e), the set of p′-elements of G is contained in Op′(G) and contains all elements of Op′(G), so it equals Op′(G) and generates it.

F2F7assume-hyp
2.1

Every g∈G therefore has a unique expression g=uk with u∈P, k∈K: existence is step 1.2, and uniqueness follows from P∩K={1}, since uk=u′k′ gives u′−1u=k′k−1∈P∩K. Defining θ(g):=u, one has θ(g1g2)=θ(g1)θ(g2) for all g1,g2: writing gi=uiki and using k1u2=u2(u2−1k1u2) with u2−1k1u2∈K, the product is (u1u2)(u2−1k1u2 k2). Thus θ:G→P is a homomorphism, it is onto because θ(u)=u, and its kernel is K.

F1F6step 1.2algebra
2.2

(c) implies (d): let θ:G→Q be an epimorphism onto a Sylow Q of G. By [F5], ∣G∣=∣ker⁡θ∣⋅∣Q∣, so ∣ker⁡θ∣=∣G∣/∣Q∣=[G:P], a p′-number by [F3]. For a p′-element x∈G, [F7] makes ord⁡(θ(x)) a divisor of ord⁡(x), hence prime to p, and also a divisor of ∣Q∣=pa, hence a power of p; so ord⁡(θ(x))=1 and θ(x)=e by [F8] and [F7]. Thus every p′-element of G lies in ker⁡θ, and since ker⁡θ is a p′-group, any product of two p′-elements lies in ker⁡θ and has order dividing ∣ker⁡θ∣, hence is a p′-element by [F7] and [F8].

F3F5F7F8step 1.2
2.3

Consequently ∣P∩Op′(G)∣=1: the order of the intersection divides both pa and the p′-number m by [F8], so ∣P Op′(G)∣=∣P∣ ∣Op′(G)∣=pam=∣G∣ by [F6], and P Op′(G)=G. Since Op′(G)⊴G by [F2] and [G:Op′(G)]=∣G∣/m=pa is a power of p, Op′(G) is a normal p-complement of G: (a) holds.

F1F2F6F8step 1.4
2.4

The moreover clauses. Assume (a)–(e) hold. Any normal p-complement K is a normal p′-subgroup, so K≤Op′(G) by [F2]; and ∣K∣=∣G∣/[G:K]=m=∣Op′(G)∣ by [F1], [F3] and [F8], because [G:K] is a p-power dividing ∣G∣=pam and ∣K∣ is a p′-number. Hence K=Op′(G), so Op′(G) is the unique normal p-complement.

F1F2F3F8step 1.4
3.1

Finally let θ:G→Q be an epimorphism as in (c). By step 2.2 every p′-element lies in ker⁡θ, while ∣ker⁡θ∣=[G:P]=m is a p′-number; so ker⁡θ is a normal p′-subgroup with [G:ker⁡θ]=∣P∣ a p-power, hence a normal p-complement, hence equal to Op′(G) by step 2.4. ∎

F1F5step 2.2step 2.4
DefinitionDefinition: Literature-sourcedProof: Not applicableaudited 2026-09-27Open item page →

P residual of a finite group

Definition

Let G be a finite group and let p be a prime. A normal subgroup N⊴G is p-cofinal when the quotient G/N is a finite p-group, that is, when [G:N] is a power of p (The quotient group G/N and coset product (gN)(hN)=ghN, If [G:N] is finite then ∣G/N∣=[G:N]; for finite G this equals ∣G∣/∣N∣, A finite p-group has order pn for a prime p and some n∈N). The p-residual of G is

Op(G):=⋂{ N⊴G:N is p-cofinal },

the intersection of all p-cofinal normal subgroups of G. It is the unique smallest normal subgroup of G whose quotient is a p-group: it is p-cofinal itself, and Op(G)≤N for every p-cofinal N.

Why the definition is well posed. The family N={N⊴G:G/N is a p-group} is nonempty (G∈N, since G/G is the trivial group, of order p0), and it is finite, because every member is a subset of the finite set G and a finite set has finitely many subsets (The cardinality ∣A∣ of a finite set, ∣P(A)∣=2∣A∣ for finite A). The intersection Op(G) is a subgroup of G by The intersection of a nonempty family of subgroups of G is a subgroup of G, and a normal subgroup because each N is normal (Normal subgroup: invariance under conjugation); it remains to see that it is again p-cofinal, and least.

Finite intersections of p-cofinal subgroups are p-cofinal. Let N1,…,Nr be p-cofinal. The diagonal map δ:G→G/N1×⋯×G/Nr, δ(g)=(gN1,…,gNr), is a homomorphism of groups (Monoid homomorphism and group homomorphism); its kernel is N1∩⋯∩Nr, and its image is a subgroup of the direct product (First isomorphism theorem for groups: G/ker⁡f≅im⁡f, The image of a group homomorphism is a subgroup and its kernel is a normal subgroup). The direct product has order ∣G/N1∣⋯∣G/Nr∣, a power of p (For finite groups G and H, ∣G×H∣=∣G∣ ∣H∣), hence is a finite p-group, and a subgroup of a finite p-group is a finite p-group (Every subgroup of a finite p-group has order a power of p). By the first isomorphism theorem G/(N1∩⋯∩Nr)≅im⁡δ, so the intersection is p-cofinal.

The intersection of all of them is a member. Since N is finite, say N={N1,…,Nr}, the preceding paragraph shows that Op(G)=N1∩⋯∩Nr is p-cofinal; in particular G/Op(G) is a p-group and Op(G)≤N for every N∈N by construction.

Smallest and unique. If K⊴G has G/K a p-group, then K∈N, so Op(G)≤K; and Op(G) itself has p-group quotient. Hence Op(G) is the smallest normal subgroup of G with p-group quotient, and it is the only one with that property, since two such subgroups contain each other.

The construction is the lower p-series counterpart of the p′-core Op′(G) of The p-prime core of a finite group: the p′-core is the largest normal p′-subgroup, while the p-residual is the smallest normal subgroup with p-group quotient. In particular Op(G) is the kernel of the natural map onto the largest p-group quotient of G, so every homomorphism from G to a finite p-group factors through G/Op(G) (The quotient group G/N and coset product (gN)(hN)=ghN).

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Sylow subgroups of a normal subgroup are intersections with Sylow subgroups

Statement

Let G be a finite group, p a prime, K⊴G a normal subgroup and P∈Syl⁡p(G) a Sylow p-subgroup (Sylow p-subgroups of a finite group). Then K∩P is a Sylow p-subgroup of K. If in addition [G:K] is a power of p, then KP=G.

Facts & Assumptions

Given: A finite group G, a prime p, a normal subgroup K⊴G, and a Sylow p-subgroup P≤G.

[F1]

Write ∣G∣=pam and ∣K∣=pcmK with p∤m, p∤mK; the p-adic valuations give c≤a, and P has order pa while a Sylow p-subgroup of K has order pc (Sylow p-subgroups of a finite group, The p-adic valuation vp(a) of a nonzero integer: the greatest k∈N with pk∣a, Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G).

[F2]

If S≤G is a p-subgroup, then S≤Q for some Sylow p-subgroup Q of G; any two Sylow p-subgroups of G are conjugate, Q=gPg−1 for some g∈G; and G has a Sylow p-subgroup (Sylow I: every finite group has a Sylow p-subgroup, Sylow II: in a finite group every p-subgroup lies in a conjugate of any Sylow p-subgroup, and the Sylow p-subgroups form a single conjugacy class).

[F3]

K is normal: gKg−1=K for every g∈G, and conjugation x↦gxg−1 is an automorphism, so ∣gSg−1∣=∣S∣ for every subgroup S (Normal subgroup: invariance under conjugation, Conjugation x↦gxg−1 is an automorphism).

[F4]

A subgroup of a finite p-group is a finite p-group, so its order is a power of p (Every subgroup of a finite p-group has order a power of p, A finite p-group has order pn for a prime p and some n∈N).

[F5]

If H≤G then ∣H∣ divides ∣G∣, and KP is a subgroup with ∣KP∣=∣K∣ ∣P∣/∣K∩P∣ (Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G, If H≤G and N⊴G, then HN is a subgroup and H∩N⊴H).

Proof

technique · direct
1.1

By [F2] applied inside the finite group K, there is a Sylow p-subgroup S of K, of order pc by [F1]; S is a p-subgroup of G, so by [F2] there is a Sylow Q of G with S≤Q, and Q=gPg−1 for some g∈G.

F1F2
1.2

On the other hand K∩P≤P, so K∩P is a finite p-group by [F4]; its order divides ∣K∣ by [F5], hence is a power of p dividing pcmK with p∤mK by [F1], and therefore ∣K∩P∣ divides pc.

F1F4F5
2.1

Then Sg−1=g−1Sg is a subgroup of K, because S≤K and K⊴G, and it has order ∣S∣=pc by [F3]; also Sg−1≤g−1Qg=P. Hence Sg−1≤K∩P, and ∣K∩P∣≥pc.

F3step 1.1
3.1

Combining steps 2.1 and 1.2, ∣K∩P∣=pc, the order of a Sylow p-subgroup of K; hence K∩P∈Syl⁡p(K), the first assertion.

F1step 2.1step 1.2
4.1

Suppose now that [G:K]=pr for some r≥0. Then ∣K∣=∣G∣/pr=pa−rm by [F1] and [F5], so the p-part of ∣K∣ is pa−r; by step 3.1, ∣K∩P∣=pa−r.

F1F5step 3.1algebra
5.1

Since KP is a subgroup of G by [F5], its order ∣K∣ ∣P∣/∣K∩P∣=pa−rm⋅pa/pa−r=pam=∣G∣ by [F5] and step 4.1; a subgroup of G with as many elements as G is G itself, so KP=G. ∎

F5step 4.1algebra
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-09-27Open item page →

P residual is generated by p prime elements and idempotent

Statement

Let G be a finite group and let p be a prime. Then:

  1. Op(G) is generated by the set of p′-elements of G;
  2. G=P Op(G) for every Sylow p-subgroup P of G;
  3. Op(Op(G))=Op(G).

Facts & Assumptions

Given: A finite group G, a prime p, and the p-residual Op(G) of P residual of a finite group; p′-elements are as in The p-prime core of a finite group.

[F1]

Op(G)⊴G, G/Op(G) is a finite p-group, and Op(G) is the least normal subgroup of G with p-group quotient (P residual of a finite group, Normal subgroup: invariance under conjugation).

[F3]

By Sylow subgroups of a normal subgroup are intersections with Sylow subgroups applied to the normal subgroup Op(G) and a Sylow P of G: P∩Op(G)∈Syl⁡p(Op(G)) and G=P Op(G), since [G:Op(G)] is a power of p by [F1] (Sylow I: every finite group has a Sylow p-subgroup, Sylow p-subgroups of a finite group).

[F4]

If a prime ℓ divides the order of a finite group H, then H has an element of order ℓ (Cauchy's theorem: if a prime p divides ∣G∣, then G has an element of order p).

Proof

technique · direct
1.1

Let Y be the set of p′-elements of G and let R:=⟨Y⟩. Then R⊴G: if y∈Y then every conjugate gyg−1 is again a p′-element by [F7], so gYg−1=Y and hence gRg−1=R by [F5].

F5F7algebra
1.2

G/R is a p-group. Suppose not; then by [F6] some prime q≠p divides [G:R], so G/R has an element yR of order q by [F4]. Let ord⁡(y)=pse with p∤e ([F1] of The p-adic valuation vp(a) of a nonzero integer: the greatest k∈N with pk∣a); then (yps)e=yord⁡(y)=e, so yps has order dividing e and is therefore a p′-element, that is, yps∈Y⊆R by [F6]. Hence (yR)ps=R, so the order of yR divides ps; but that order is q, a prime different from p, a contradiction.

F4F5F6algebra
1.3

Claim 2 is exactly the second assertion of [F3].

F3
2.1

Therefore Op(G)≤R: R is a normal subgroup of G with p-group quotient by steps 1.1 and 1.2, and Op(G) is the least such by [F1].

F1step 1.1step 1.2
3.1

Conversely Y⊆Op(G): the natural map G→G/Op(G) is a homomorphism onto a finite p-group by [F1], so it kills every p′-element by [F2]. Hence R=⟨Y⟩⊆Op(G), and with step 2.1, R=Op(G); this is claim 1.

F1F2step 2.1
4.1

Claim 3: put K:=Op(G). By claim 1 applied to the finite group K, Op(K) is generated by the p′-elements of K; every such element is a p′-element of G and so lies in K trivially, while conversely claim 1 gives K=⟨Y⟩ with Y the p′-elements of G, and every y∈Y is an element of K and hence a p′-element of K. The two generating sets agree, so Op(K)=K. ∎

step 3.1algebra
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Abelian sylow fusion in its normalizer

Statement

Let G be a finite group, p a prime and P∈Syl⁡p(G) an abelian Sylow p-subgroup. If x,y∈P are conjugate in G, then x and y are conjugate in NG(P): there is u∈NG(P) with y=uxu−1.

Facts & Assumptions

Given: A finite group G, a prime p, an abelian P∈Syl⁡p(G), elements x,y∈P and g∈G with y=gxg−1.

[F1]

Write ∣G∣=pam with p∤m; then ∣P∣=pa, and a subgroup H≤G is a Sylow p-subgroup of G exactly when ∣H∣=pa; a p-subgroup of G of order pa is a Sylow p-subgroup (Sylow p-subgroups of a finite group, Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G).

[F2]

Every p-subgroup of G is contained in a Sylow p-subgroup of G, and any two Sylow p-subgroups of G are conjugate: for every Sylow Q there is u∈G with Q=uPu−1 (Sylow II: in a finite group every p-subgroup lies in a conjugate of any Sylow p-subgroup, and the Sylow p-subgroups form a single conjugacy class).

[F3]

As P is abelian, every element of P commutes with x and with y; that is, P≤CG(y) and P≤CG(x), where CG(y)={u∈G:uy=yu} (The centralizer CG(H) of a subgroup, The conjugacy class Cl⁡G(x) and centralizer CG(x) of an element).

[F4]

For every u∈G the map cu(z)=uzu−1 is an automorphism of G, so cu(zw)=cu(z)cu(w), cu(z)−1=cu(z−1) and cu is injective; also u∈NG(P) exactly when uPu−1=P (Conjugation x↦gxg−1 is an automorphism, The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup).

[F5]

CG(y) and NG(P) are subgroups of G (CG(x) and NG(H) are subgroups of G).

Proof

technique · direct
1.1

Pg=gPg−1 is contained in CG(y): for w∈P one has z:=gwg−1∈Pg and y=gxg−1, so zy=gwg−1gxg−1=g(wx)g−1=g(xw)g−1=yz by [F4] and the commutativity of P in [F3]. Also P≤CG(y) by [F3].

F3F4given
2.1

Both P and Pg are Sylow p-subgroups of CG(y): each has order pa by [F1] and [F4], and each is contained in CG(y) by step 1.1; since ∣CG(y)∣ divides ∣G∣=pam by [F1], the exact power of p dividing ∣CG(y)∣ is pa, so a subgroup of CG(y) of order pa is a Sylow p-subgroup of CG(y) by [F1] applied to CG(y).

F1F2F5step 1.1
3.1

By step 2.1 and Sylow conjugacy inside the finite group CG(y) [F2], there is c∈CG(y) with (Pg)c=P; since (Pg)c=c(gPg−1)c−1=(cg)P(cg)−1, [F4] gives cg∈NG(P).

F2F4step 2.1
4.1

For this c one has (xg)c=yc=y, the first equality by [F4] applied twice and the second because c∈CG(y).

F4step 3.1
5.1

Since (xg)c=xcg=y by step 4.1, x and y are conjugate by the element cg∈NG(P), as claimed. ∎

step 3.1step 4.1
TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Burnside normal p complement theorem

Statement

Let G be a finite group, p a prime and P∈Syl⁡p(G). If P≤Z(NG(P)), that is, if every element of P commutes with every element of the normalizer NG(P) (The center Z(G) of a group, The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup), then G has a normal p-complement.

Facts & Assumptions

Given: A finite group G, a prime p, a Sylow p-subgroup P≤G with P≤Z(NG(P)).

[F1]

Write ∣G∣=pam with p∤m; then ∣P∣=pa and the index n:=[G:P]=m is prime to p (Sylow p-subgroups of a finite group, Sylow I: every finite group has a Sylow p-subgroup, Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G).

[F2]

As P≤Z(NG(P)) and P≤NG(P), every two elements of P commute: P is abelian (The center Z(G) of a group, The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup).

[F3]

The transfer V=Vφ of the homomorphism φ=id⁡P:P→P is a homomorphism G→P, explicitly V(x)=∏α∈P\Gφ(tαxtαx−1) for any transversal, and it agrees with the cycle formula V(x)=∏iφ(tixniti−1) where ni are the orbit sizes of ⟨x⟩ acting on P\G (Transfer homomorphism for a finite index subgroup, Transfer is a homomorphism, Transfer cycle decomposition formula, Transfer is independent of the transversal).

[F4]

The orbits of the action of ⟨x⟩ on the finite set P\G partition it, so their sizes n1,…,nr satisfy n1+⋯+nr=[G:P]=n (The orbits of a group action are the equivalence classes of x∼y iff y=g⋅x for some g, and hence partition the acted-on set, Left group actions, transitive actions, and faithful actions).

[F5]

If P is abelian and u,v∈P are conjugate in G, then they are conjugate in NG(P) (Abelian sylow fusion in its normalizer).

[F7]

If a finite group has a Sylow p-subgroup Q and an epimorphism θ:G→Q, then it has a normal p-complement (Equivalent forms of having a normal p complement).

Proof

technique · direct
1.1

By [F2] the group P is abelian, so the identity map φ:P→P is a homomorphism into an abelian group and the transfer V:G→P of [F3] is defined; fix a transversal and let ni be the orbit sizes of ⟨x⟩ on P\G.

F2F3
1.2

For x∈P, the cycle formula of [F3] gives V(x)=∏iφ(tixniti−1)=∏itixniti−1, each factor lying in P; also xni∈P and ∑ini=n by [F4].

F3F4
2.1

For each i the element tixniti−1 equals (xni)ti, so it is a G-conjugate of xni, and both lie in P; by [F5] there is ui∈NG(P) with tixniti−1=(xni)ui.

F5F8step 1.2
3.1

Since xni∈P≤Z(NG(P)) commutes with ui∈NG(P), step 2.1 gives tixniti−1=xni. Hence V(x)=∏ixni=xn1+⋯+nr=xn by [F4], all factors being powers of x (Exponent laws in a group: gm+n=gmgn and (gm)n=gmn for all m,n∈Z, and (gh)n=gnhn when g and h commute).

F4step 1.2step 2.1algebra
4.1

The power map y↦yn is a bijection P→P by [F6], since gcd⁡(n,∣P∣)=1 by [F1]; therefore V(P)={xn:x∈P}=P by step 3.1, and V is an epimorphism G→P.

F1F6step 3.1
5.1

Applying [F7] with Q:=P and θ:=V yields that G has a normal p-complement. ∎

F7step 4.1
DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-6-sol)audited 2026-09-27Open item page →

P local normalizer for normal complement theory

Definition

Let G be a finite group, p a prime, and P∈Syl⁡p(G) a Sylow p-subgroup (Sylow p-subgroups of a finite group). On this page a nontrivial p-local normalizer of G means a subgroup of the form

NG(Q)={g∈G:gQg−1=Q},

where Q is a nontrivial subgroup of P, that is 1≠Q≤P (The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup, Subgroup). The quantifier always excludes Q=1: since NG(1)=G, admitting the trivial subgroup would make the local statements below vacuous or false.

Why normalizers suffice up to conjugacy. Every nontrivial p-subgroup S of G is contained in a Sylow p-subgroup of G, all of which are conjugate to P; so there is g∈G with Sg=gSg−1≤P and NG(Sg)=NG(S)g, by the conjugation automorphism Conjugation x↦gxg−1 is an automorphism (Sylow II: in a finite group every p-subgroup lies in a conjugate of any Sylow p-subgroup, and the Sylow p-subgroups form a single conjugacy class). Thus the family {NG(Q):1≠Q≤P} represents every normalizer of a nontrivial p-subgroup of G up to conjugacy, and the claim "NG(Q) has a normal p-complement for every 1≠Q≤P" is invariant under replacing P by a conjugate Sylow subgroup. The normalizers are subgroups of G by CG(x) and NG(H) are subgroups of G.

Centralizers are named separately. The broader convention in the literature calls both the normalizers NG(Q) and the centralizers CG(Q) of nontrivial p-subgroups p-local subgroups. The two theorems of this page that are stated locally — the inheritance lemma Normal p complements pass to subgroups and p local normalizers and Frobenius' normal p-complement theorem Frobenius normal p complement theorem — quantify over the normalizers NG(Q) only, so that is the meaning fixed here; centralizers (The centralizer CG(H) of a subgroup) are never silently included. A normalizer of a p-subgroup is itself a subgroup whose Sylow p-subgroups are again to be read with Sylow p-subgroups of a finite group and A finite p-group has order pn for a prime p and some n∈N.

DefinitionDefinition: Literature-sourcedProof: Not applicableaudited 2026-09-27Open item page →

Control of fusion in a sylow p subgroup

Definition

Let G be a finite group, p a prime, and P∈Syl⁡p(G) a Sylow p-subgroup (Sylow p-subgroups of a finite group). Following the convention xg=gxg−1 of The conjugacy class Cl⁡G(x) and centralizer CG(x) of an element, we say that

P controls fusion in P with respect to G

when, for all x,y∈P, the existence of g∈G with y=xg implies the existence of u∈P with y=xu. Equivalently: any two elements of P that are G-conjugate are already conjugate by an element of P; equivalently, the conjugacy class of x in G meets P in exactly the conjugacy class of x under P (Subgroup, The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup).

What is and is not claimed. This is a statement about element fusion only: it says nothing about when two subgroups of P are conjugate in G, and nothing about elements of G outside P. It is a property of the pair (G,P), and it is invariant under conjugating P: if g∈G and the Sylow P controls fusion in P with respect to G, then Pg=gPg−1 controls fusion in Pg with respect to G, because y=xu implies yg=(xg)gug−1 and gug−1∈Pg whenever u∈P. Control by P implies control by NG(P), since P≤NG(P): every conjugating element supplied inside P also belongs to NG(P). When NG(P)=P the two conditions are identical, but this equality alone does not assert that either condition holds.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-09-27Open item page →

Proper subgroup of a finite p group is properly normalized local

Statement

Let P be a finite p-group and let S<P be a proper subgroup. Then S<NP(S): the normalizer of S in P strictly contains S.

Facts & Assumptions

Given: A prime p, a finite p-group P, and a proper subgroup S<P; the assertion is proved for all finite p-groups of order <∣P∣ (induction hypothesis).

[F1]

S<NP(S) means that NP(S) is a subgroup of P containing S properly, i.e. that there is x∈P with xSx−1=S and x∉S (The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup, Subgroup).

[F2]

Every subgroup of P is a finite p-group, so ∣S∣ is a power of p; if P≠1 then ∣P∣=pr with r≥1; S=P if and only if ∣S∣=∣P∣ (Every subgroup of a finite p-group has order a power of p, A finite p-group has order pn for a prime p and some n∈N, Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G).

[F3]

If P≠1 then Z(P)≠1; Z(P)⊴P; and Z(P) consists of the elements commuting with every element of P, so zSz−1=S for every z∈Z(P) (Every nontrivial finite p-group has nontrivial center, in fact p divides ∣Z(P)∣, The center Z(G) of a group, The center of a group is a normal subgroup, Normal subgroup: invariance under conjugation, Conjugation x↦gxg−1 is an automorphism).

[F5]

For Z⊴P and S≤P the image SZ/Z is a subgroup of P/Z, and if Z≤S then SZ/Z=S/Z with ∣S/Z∣=∣S∣/∣Z∣; moreover conjugation x↦uxu−1 is an automorphism, so ∣xSx−1∣=∣S∣ and xSx−1Z is the image of xSx−1 in P/Z (The image of a group homomorphism is a subgroup and its kernel is a normal subgroup, Monoid homomorphism and group homomorphism, If H≤G and N⊴G, then HN is a subgroup and H∩N⊴H, If [G:N] is finite then ∣G/N∣=[G:N]; for finite G this equals ∣G∣/∣N∣, Conjugation x↦gxg−1 is an automorphism).

[F6]

Strong induction on the natural number ∣P∣: if, for every N, truth for all finite p-groups of order <N implies truth for all of order N, then the statement holds for all finite p-groups (Strong (complete) induction).

Proof

technique · direct
1.1

The case P=1 is vacuous, since it has no proper subgroup. Assume the assertion known for every finite p-group of order <∣P∣.

F2F6given
1.2

If S=1 then every x∈P satisfies xSx−1=S, so NP(S)=P; if also S<P then P≠1 and NP(S)=P>S, which is the claim.

F1F3given
1.3

Suppose Z(P)⊈S: choose z∈Z(P) with z∉S. By [F3] z normalizes S, so z∈NP(S)∖S and S<NP(S).

F1F3choose
1.4

It remains to treat the case Z≤S with Z:=Z(P), under S≠1 and P≠1. Then P/Z is a finite p-group of order ∣P∣/∣Z∣<∣P∣ by [F3] and [F4], and T:=S/Z is a subgroup of it by [F5]. If T=P/Z then S=P by [F5], contrary to hypothesis, so T<P/Z.

F2F3F4F5given
2.1

The induction hypothesis of step 1.1 applies to the finite p-group P/Z and its proper subgroup T: there is a coset xZ∈NP/Z(T) with xZ∉T. By the definition of the normalizer this means (xZ)T(xZ)−1=T in P/Z.

F1F4F5step 1.4assume-hyp
3.1

Translating back: (xSx−1)Z/Z=SZ/Z. Since Z≤S, this says xSx−1Z=S, so every element of xSx−1 lies in S; thus xSx−1⊆S, and since conjugation is injective with ∣xSx−1∣=∣S∣, actually xSx−1=S. Hence x∈NP(S) by [F1].

F5step 2.1algebra
3.2

Moreover x∉S: otherwise xZ∈S/Z=T, contrary to the choice in step 2.1.

F5step 2.1
4.1

So in the case Z(P)≤S there is x∈NP(S)∖S as well, and together with steps 1.2 and 1.3 this proves S<NP(S) in every case, completing the induction. ∎

F1step 1.2step 1.3step 3.1step 3.2
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Normal p complements pass to subgroups and p local normalizers

Statement

Let G be a finite group, p a prime, and suppose G has a normal p-complement K (Normal p complement and p nilpotent group). Then:

  1. every subgroup H≤G has a normal p-complement, namely K∩H;
  2. every quotient G/L by a normal subgroup L⊴G has a normal p-complement, namely KL/L;
  3. in particular NG(Q) has a normal p-complement for every nontrivial p-subgroup Q≤G (P local normalizer for normal complement theory).

Facts & Assumptions

Given: A finite group G, a prime p, a normal p-complement K⊴G, a subgroup H≤G, and a normal subgroup L⊴G.

[F1]

K⊴G, p∤∣K∣, and [G:K] is a power of p; equivalently G has a normal p′-subgroup of p-power index, and a finite group has a normal p-complement exactly when it has such a subgroup (Normal p complement and p nilpotent group, Equivalent forms of having a normal p complement).

[F2]

If K⊴G and H≤G, then H∩K⊴H, HK≤G and H/(H∩K)≅HK/K, so [H:H∩K]=∣HK/K∣ (Second isomorphism theorem for groups: H/(H∩N)≅HN/N, If H≤G and N⊴G, then HN is a subgroup and H∩N⊴H, Normal subgroup: invariance under conjugation).

[F3]

Orders divide: if X≤Y then ∣X∣ divides ∣Y∣, and ∣Y∣=∣X∣ [Y:X]; a group whose order divides a power of p is a finite p-group (Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G, A finite p-group has order pn for a prime p and some n∈N, Every subgroup of a finite p-group has order a power of p).

[F4]

If L⊴G with L≤KL, then KL⊴G and (G/L)/(KL/L)≅G/KL; the quotient KL/L is the image of K under the natural map G→G/L, and ∣KL/L∣=∣K∣/∣K∩L∣ (Third isomorphism theorem for groups: (G/K)/(N/K)≅G/N, The quotient group G/N and coset product (gN)(hN)=ghN, If [G:N] is finite then ∣G/N∣=[G:N]; for finite G this equals ∣G∣/∣N∣, Second isomorphism theorem for groups: H/(H∩N)≅HN/N).

[F6]

A normalizer NG(Q) of a nontrivial p-subgroup Q is a subgroup of G (The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup, Subgroup).

Proof

technique · direct
1.1

K∩H⊴H by [F2], and ∣K∩H∣ divides ∣K∣ by [F3], so p∤∣K∩H∣.

F2F3
1.2

Moreover [H:H∩K]=∣HK/K∣ by [F2], and HK/K≤G/K, so [H:H∩K] divides [G:K] by [F3] and is a power of p. Hence K∩H is a normal p′-subgroup of H of p-power index, i.e. a normal p-complement of H by [F1].

F1F2F3
1.3

KL⊴G and KL/L⊴G/L by [F4]; and ∣KL/L∣=∣K∣/∣K∩L∣ divides ∣K∣, so p∤∣KL/L∣.

F3F4
1.4

By [F4] and [F5], ∣(G/L):(KL/L)∣=∣G/KL∣ divides ∣G/K∣, a power of p; hence (G/L):(KL/L) is a power of p and KL/L is a normal p-complement of G/L by [F1].

F1F4F5
2.1

For a nontrivial p-subgroup Q≤G the normalizer H:=NG(Q) is a subgroup of G by [F6], so step 1.2 gives that K∩NG(Q) is a normal p-complement of NG(Q); with steps 1.1 and 1.4 this establishes all three assertions. ∎

F6step 1.1step 1.2step 1.4
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Fusion control and centralizer transitivity are equivalent

Statement

Let H be a finite group, p a prime, and P∈Syl⁡p(H) (Sylow p-subgroups of a finite group). The following are equivalent.

(i) NH(P) controls fusion in P with respect to H: whenever x,y∈P and y=hxh−1 for some h∈H, there is u∈NH(P) with y=uxu−1. (ii) For every x∈P with x≠e, the centralizer CH(x)={h∈H:hx=xh} acts by conjugation transitively on the set Syl⁡p(H;x):={T∈Syl⁡p(H):x∈T} of Sylow p-subgroups of H containing x; that is, for any T1,T2∈Syl⁡p(H;x) there is c∈CH(x) with T2=T1c=cT1c−1.

Facts & Assumptions

Given: A finite group H, a prime p, a Sylow p-subgroup P≤H, and the notation xh=hxh−1 of The conjugacy class Cl⁡G(x) and centralizer CG(x) of an element.

[F1]

Sylow p-subgroups of H are conjugate, and the conjugate of a Sylow p-subgroup by any element of H is again a Sylow p-subgroup (Sylow II: in a finite group every p-subgroup lies in a conjugate of any Sylow p-subgroup, and the Sylow p-subgroups form a single conjugacy class, Sylow p-subgroups of a finite group).

[F3]

CH(x) and NH(P) are subgroups of H; z∈CH(x) satisfies xz=x, and u∈NH(P) satisfies Pu=P (CG(x) and NG(H) are subgroups of G, The centralizer CG(H) of a subgroup, The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup, Subgroup).

[F4]

If T∈Syl⁡p(H), x∈T and h∈H, then Th∈Syl⁡p(H) and xh∈Th; hence conjugation by h carries Syl⁡p(H;x) into Syl⁡p(H;xh), and if c∈CH(x) then Tc∈Syl⁡p(H;x) for every T∈Syl⁡p(H;x) (Sylow II: in a finite group every p-subgroup lies in a conjugate of any Sylow p-subgroup, and the Sylow p-subgroups form a single conjugacy class, Sylow p-subgroups of a finite group, Conjugation x↦gxg−1 is an automorphism).

Proof

technique · direct
1.1

(ii) implies (i). Assume (ii), and let x,y∈P, h∈H with y=xh=hxh−1. If x=e then y=e=ee with e∈NH(P), so assume x≠e. Then x=yh−1 by [F2], so x∈P∩Ph−1; by [F1] and [F4] both P and Ph−1 lie in Syl⁡p(H;x), so (ii) provides c∈CH(x) with Ph−1=Pc.

F1F2F4assume-hyp
1.2

(i) implies (ii). Assume (i). Let x∈P with x≠e and let T∈Syl⁡p(H;x). By [F1] there is g∈H with T=Pg; then x∈Pg gives a:=xg−1=g−1xg∈P, and ag=x by [F2], so a and x are H-conjugate elements of P and (i) provides u∈NH(P) with au=x.

F1F2assume-hyp
2.1

The equality Ph−1=Pc says h−1Ph=cPc−1. Multiplying on the left by h and on the right by h−1 gives P=(hc)P(hc)−1, so n:=hc satisfies Pn=P, that is n∈NH(P).

F2F3step 1.1
3.1

Moreover xn=(hc)x(hc)−1=h(cxc−1)h−1=hxh−1=y, since c centralizes x. So y is conjugate to x by the element n∈NH(P), which proves (i).

F3step 2.1
4.1

Put v:=ug−1, so that v−1=gu−1 and g=v−1u. The identity au=x reads u(g−1xg)u−1=x, that is xv=x by [F2]; hence v∈CH(x). Moreover T=Pg=Pv−1u=(Pu)v−1=Pv−1 by [F2], since u∈NH(P); and v−1∈CH(x). So every member T of Syl⁡p(H;x) equals Pc for the element c:=v−1∈CH(x), which is (ii). ∎

F2F3step 1.2
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-09-27Open item page →

Local sylow conjugacy ascent for fusion

Statement

Let G be a finite group, p a prime and P∈Syl⁡p(G). Suppose that for every nontrivial p-subgroup S≤P and every x∈S with x≠e, the centralizer CNG(S)(x) acts transitively on the Sylow p-subgroups of NG(S) containing x (P local normalizer for normal complement theory). Then NG(P) controls fusion in P with respect to G: any two G-conjugate elements of P are conjugate by an element of NG(P) (Control of fusion in a sylow p subgroup).

Facts & Assumptions

Given: A finite group G, a prime p, a Sylow p-subgroup P≤G, and the hypothesis that for every nontrivial p-subgroup S≤P and every x∈S∖{e}, the centralizer CNG(S)(x) acts transitively on the Sylow p-subgroups of NG(S) containing x.

[F2]

The local hypothesis says that for every nontrivial p-subgroup S≤P, every x∈S∖{e} and all Sylow p-subgroups T1,T2 of NG(S) containing x, there is c∈CNG(S)(x) with T2=T1c (P local normalizer for normal complement theory).

[F4]

Conjugation z↦gzg−1 is an automorphism, (za)b=zba, equivalently zab=(zb)a; CG(x) and all normalizers are subgroups; v∈NG(T) gives Tv=T, and v∈CG(x) gives xv=x; also S≤NG(S) (Conjugation x↦gxg−1 is an automorphism, In a group e−1=e, (g−1)−1=g and (gh)−1=h−1g−1, the order of the last product being essential, CG(x) and NG(H) are subgroups of G, The centralizer CG(H) of a subgroup, The conjugacy class Cl⁡G(x) and centralizer CG(x) of an element, Subgroup).

[F5]

Orders: all Sylow p-subgroups of a finite group X have the same order, equal to the exact power of p dividing ∣X∣; a subgroup's order divides the group's order (Sylow p-subgroups of a finite group, Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G, Every subgroup of a finite p-group has order a power of p).

[F6]

Strong induction on the positive integer m(Q,R):=∣P∣/∣Q∩R∣, for Sylow p-subgroups Q,R of G: ∣Q∩R∣ divides ∣Q∣=∣P∣ by [F5], so m(Q,R) is a positive integer, and m(Q,R)=1 exactly when Q=R (Strong (complete) induction, Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G).

Proof

technique · direct
1.1

Let x∈P, x≠e, and let Ω be the set of Sylow p-subgroups of G containing x. We prove by strong induction on m(Q,R) that any two members Q,R∈Ω are conjugate by an element of CG(x); the case m(Q,R)=1, that is Q=R, is trivial by [F6].

F6given
1.2

The hypothesis extends to every nontrivial p-subgroup of G, not only to those inside P: let S≤G be a nontrivial p-subgroup and choose d∈G with Sd≤P by [F1], so that NG(Sd)=NG(S)d and conjugation by d carries Sylow p-subgroups of NG(S) to Sylow p-subgroups of NG(Sd). If x∈S∖{e} and T1,T2 are Sylow p-subgroups of NG(S) containing x, then xd∈Sd∖{e} and T1d,T2d are Sylow p-subgroups of NG(Sd) containing xd, so [F2] applied to the nontrivial p-subgroup Sd≤P provides c′∈CNG(Sd)(xd) with (T1d)c′=T2d. Then c:=c′d−1 normalizes S and centralizes x, so c∈CNG(S)(x), and conjugating the displayed equality by d−1 gives T1c=T2; hence the form [F2] of the hypothesis holds for S.

F1F2F4algebra
2.1

For the induction step let Q,R∈Ω with S:=Q∩R and m(Q,R)>1, so that S<Q and S<R by [F6]; note x∈S, so S is a nontrivial p-subgroup, being a subgroup of the p-group Q by [F5]. By [F3] there are NQ(S) and NR(S) with S<NQ(S)≤Q and S<NR(S)≤R.

F3F5F6step 1.1
3.1

Put N:=NG(S), which is a p-local normalizer; by [F1] there is a Sylow p-subgroup TQ of N with NQ(S)≤TQ and a Sylow p-subgroup TR of N with NR(S)≤TR; then x∈S<NQ(S)≤TQ and x∈S<NR(S)≤TR by [F4]. By [F2] in the form of step 1.2, applied to the nontrivial p-subgroup S of step 2.1, the element x∈S and the Sylows TQ,TR of NG(S) containing x, there is c∈CNG(S)(x) with TQc=TR.

F1F2F4step 2.1step 1.2
4.1

By [F1] choose Sylow p-subgroups Q∗,R∗ of G with TQ≤Q∗ and TR≤R∗. Then x∈TQ≤Q∗ and x∈TR≤R∗, so Q∗,R∗∈Ω. Moreover S<NQ(S)≤Q∩Q∗ and S<NR(S)≤R∩R∗, and since TR=TQc≤(Q∗)c also TR≤(Q∗)c∩R∗ with ∣TR∣≥∣NR(S)∣>∣S∣. Hence m(Q,Q∗)<m(Q,R), m(R,R∗)<m(Q,R) and m((Q∗)c,R∗)<m(Q,R) by [F5] and [F6].

F1F5F6step 2.1step 3.1
5.1

Also (Q∗)c∈Ω because xc=x and x∈Q∗, and c∈CG(x), so (Q∗)c is CG(x)-conjugate to Q∗.

F4step 3.1step 4.1
6.1

The induction hypothesis of step 1.1 applies to the pairs (Q,Q∗), ((Q∗)c,R∗) and (R∗,R), all of whose measures are smaller than m(Q,R): Q is CG(x)-conjugate to Q∗, (Q∗)c to R∗, and R∗ to R. Since CG(x)-conjugacy is an equivalence relation, and since (Q∗)c is CG(x)-conjugate to Q∗ by step 5.1, the element Q is CG(x)-conjugate to R, completing the induction.

step 1.1step 4.1step 5.1algebra
7.1

We have shown that for every x∈P with x≠e, CG(x) acts transitively on the Sylow p-subgroups of G containing x; by Fusion control and centralizer transitivity are equivalent applied with H:=G, the normalizer NG(P) controls fusion in P with respect to G. ∎

step 1.1step 6.1
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-09-27Open item page →

Local normal p complements force control of fusion

Statement

Let G be a finite group, p a prime and P∈Syl⁡p(G) a Sylow p-subgroup (Sylow p-subgroups of a finite group). Suppose that every nontrivial p-local normalizer NG(Q), 1≠Q≤P, has a normal p-complement (P local normalizer for normal complement theory, Normal p complement and p nilpotent group). Then P controls fusion in P with respect to G (Control of fusion in a sylow p subgroup): whenever x,y∈P and y=xg=gxg−1 for some g∈G, there is v∈P with y=xv.

Facts & Assumptions

Given: A finite group G, a prime p, a Sylow p-subgroup P≤G, and the hypothesis that NG(Q) has a normal p-complement for every subgroup Q with 1≠Q≤P.

[F1]

Normal p-complement structure: if a finite group H has a normal p-complement K, then K⊴H, p∤∣K∣, [H:K] is a power of p, and for every Sylow p-subgroup S of H one has [H:K]=∣S∣, S∩K={1} and H=KS=SK; thus every h∈H can be written h=sk with s∈S, k∈K (Normal p complement and p nilpotent group, Sylow p-subgroups of a finite group, Normal subgroup: invariance under conjugation).

[F2]

Conjugation and commutators: xg=gxg−1, (xa)b=xba, equivalently xab=(xb)a; [a,b]=aba−1b−1; if N⊴H, a∈N and h∈H, then hah−1∈N, so [h,a]=hah−1a−1∈N as well; and [a,b], as a product of a and b, lies in every subgroup containing both a and b (The conjugacy class Cl⁡G(x) and centralizer CG(x) of an element, In a group e−1=e, (g−1)−1=g and (gh)−1=h−1g−1, the order of the last product being essential, Commutators [g,h]=ghg−1h−1 and the commutator subgroup [G,G], Normal subgroup: invariance under conjugation, Conjugation x↦gxg−1 is an automorphism, Subgroup).

[F4]

T≤NH(T) for every subgroup T≤H: every element of T normalizes T (The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup, CG(x) and NG(H) are subgroups of G).

[F5]

If a Sylow p-subgroup T of a finite group H controls fusion in T, then its normalizer also controls fusion there, and Fusion control and centralizer transitivity are equivalent says this is equivalent to CH(x) acting transitively on Sylow p-subgroups of H containing every x∈T∖{e}. Local Sylow conjugacy ascent (Local sylow conjugacy ascent for fusion) needs this centralizer transitivity only for x∈S∖{e}, for each nontrivial S≤P and H=NG(S).

[F6]

Proof

technique · direct
1.1

(A p-nilpotent group is controlled by its Sylow subgroups.) Let H be a finite group with a normal p-complement K, and let S∈Syl⁡p(H). Let x,y∈S and h∈H with y=xh. By [F1], H=KS, so we may write h=ks with k∈K, s∈S, and by [F2] y=xks=(xs)k. Put a:=xs∈S, so that y=ak=kak−1 and hence [k,a]=kak−1a−1=ya−1: this element lies in S, because y,a∈S, and it also equals k (ak−1a−1), which lies in K because ak−1a−1∈K by normality of K and k∈K. Therefore [k,a]∈S∩K={1} by [F1], so ya−1=1 and y=a=xs with s∈S. Thus every H-conjugacy between elements of S is realized inside S: S controls fusion in S with respect to H.

F1F2given
2.1

Let S≤P be a nontrivial subgroup; since P is a finite p-group, the subgroup S is a finite p-group by [F6]. Let T∈Syl⁡p(NG(S)). The hypothesis gives that N:=NG(S) has a normal p-complement, so by step 1.1 applied to H:=N and the Sylow T of N, the group T controls fusion in T with respect to N; by [F4] we have T≤NN(T), so the normalizer NNG(S)(T) controls fusion in T with respect to NG(S) as well.

F4F6givenstep 1.1
2.2

If P≠{1}, then P is a nontrivial subgroup of P, so the hypothesis gives that NG(P) has a normal p-complement; by [F3] the subgroup P is a Sylow p-subgroup of NG(P), and by step 1.1 applied to H:=NG(P) with the Sylow P, the group P controls fusion in P with respect to NG(P).

F3givenstep 1.1
3.1

Let S≤P be nontrivial and x∈S∖{e}. Since S≤NG(S), some Sylow T of NG(S) contains S and hence x. By step 2.1, T controls fusion in T with respect to NG(S), so [F5] gives centralizer transitivity for x. As S and x were arbitrary, the hypothesis of the local Sylow conjugacy ascent in [F5] is satisfied; therefore NG(P) controls fusion in P with respect to G.

F5step 2.1
4.1

Let x,y∈P and g∈G with y=xg. If P≠{1}, step 3.1 provides u∈NG(P) with y=xu, and then step 2.2 provides v∈P with y=xv. If P={1} then x=y=e and y=xe with e∈P, so P controls fusion in P in this case too. ∎

step 2.2step 3.1
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Normal p subgroup has proper commutator in a p group

Statement

Let P be a finite p-group and let Q⊴P be a nontrivial normal subgroup (A finite p-group has order pn for a prime p and some n∈N, Normal subgroup: invariance under conjugation). Then

[Q,P]<Q,

where [Q,P]=⟨[q,u]:q∈Q, u∈P⟩ is the subgroup commutator of Subgroup commutators and the lower central series and Commutators [g,h]=ghg−1h−1 and the commutator subgroup [G,G]. Moreover [Q,P]⊴Q, so the quotient Q/[Q,P] is defined (The quotient group G/N and coset product (gN)(hN)=ghN).

Facts & Assumptions

Given: A finite p-group P and a nontrivial normal subgroup Q⊴P.

[F1]

Q∩Z(P)≠{1}: a nontrivial normal subgroup of a finite p-group meets the center nontrivially (Every nontrivial normal subgroup of a finite p-group meets the center nontrivially, The center Z(G) of a group).

[F2]

Commutators: [a,b]=aba−1b−1, and [A,B]=⟨[a,b]:a∈A, b∈B⟩; if A≤B≤G and D≤G then every generator [a,d] of [A,D] is a generator of [B,D], so [A,D]≤[B,D]. If N⊴G, A≤N and B≤G, then [A,B]≤N: for a∈A, b∈B one has bab−1∈N, so [a,b]=a (bab−1)−1∈N (Subgroup commutators and the lower central series, Commutators [g,h]=ghg−1h−1 and the commutator subgroup [G,G], The subgroup ⟨S⟩ generated by a subset, the cyclic subgroup ⟨g⟩, and cyclic groups, Normal subgroup: invariance under conjugation, In a group e−1=e, (g−1)−1=g and (gh)−1=h−1g−1, the order of the last product being essential, Subgroup).

[F3]

If M⊴G, the quotient map π:G→G/M, π(g)=gM, is a surjective group homomorphism, π(A)=AM/M for every subgroup A≤G, and images of generated subgroups are generated by the images of the generators (The quotient group G/N and coset product (gN)(hN)=ghN, Homomorphisms respect commutator subgroups and derived series, The subgroup ⟨S⟩ generated by a subset, the cyclic subgroup ⟨g⟩, and cyclic groups).

[F4]
[F5]

Strong induction on the positive integer ∣Q∣ (Strong (complete) induction).

Proof

technique · direct
1.1

We prove by strong induction on n=∣Q∣ the statement: for every finite p-group P and every nontrivial Q⊴P with ∣Q∣=n, one has [Q,P]<Q and [Q,P]⊴Q. Let such P and Q be given. By [F1] the subgroup Z0:=Q∩Z(P) is nontrivial; it is normal in P as the intersection of the normal subgroups Q and Z(P) (the center is normal), and [Z0,P]={1} because every element of Z0 commutes with every element of P.

F1F2given
2.1

If Z0=Q, that is Q≤Z(P), then every generator [q,u] of [Q,P] equals 1, so [Q,P]={1}<Q since Q is nontrivial; also [Q,P]={1}⊴Q.

F2F5step 1.1
2.2

Otherwise Z0<Q, so Qˉ:=Q/Z0 is a nontrivial normal subgroup of the finite p-group Pˉ:=P/Z0 by [F4]; since ∣Qˉ∣=∣Q∣/∣Z0∣<∣Q∣=n, the induction hypothesis of step 1.1 applies to Pˉ and Qˉ and gives [Qˉ,Pˉ]<Qˉ.

F3F4F5step 1.1
3.1

Let π:P→Pˉ be the quotient map. By [F3], π(Q)=Qˉ, π(P)=Pˉ, and π([Q,P]) is generated by the elements π([q,u])=[π(q),π(u)] for q∈Q, u∈P, that is π([Q,P])=[Qˉ,Pˉ]; on the other hand π([Q,P])=[Q,P]Z0/Z0.

F2F3step 2.2
4.1

The inclusion [Qˉ,Pˉ]<Qˉ of step 2.2 therefore says [Q,P]Z0/Z0<Q/Z0, which means [Q,P]Z0<Q; as [Q,P]≤[Q,P]Z0, we get [Q,P]<Q.

F4step 2.2step 3.1
5.1

Finally [Q,P]⊴Q: for q∈Q and a∈Q, b∈P one has q[a,b]q−1=[qaq−1,qbq−1]=[aq,bq], and aq∈Q, bq∈P because q∈Q≤P and Pq=P; hence conjugation by q permutes the generators of [Q,P], so [Q,P]q=[Q,P] for every q∈Q, which is normality of [Q,P] in Q. This completes the induction and the proof. ∎

F2F3step 1.1step 4.1
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-09-27Open item page →

Fusion control forces trivial Sylow intersection with the p residual

Statement

Let G be a finite group, p a prime and P∈Syl⁡p(G) a Sylow p-subgroup (Sylow p-subgroups of a finite group), and put K:=Op(G), the p-residual (P residual of a finite group). Suppose that P controls fusion in P with respect to G (Control of fusion in a sylow p subgroup) and set Q:=P∩K. Then Q={1}: that is, P∩Op(G)={1}.

More precisely, if Q≠{1}, then the transfer V ⁣:K→Q/[Q,P] of the quotient map Q→Q/[Q,P] (Transfer homomorphism for a finite index subgroup, Transfer is a homomorphism, Subgroup commutators and the lower central series) is a nontrivial homomorphism onto a nontrivial finite abelian p-group, so Op(K)<K; since Op(Op(G))=Op(G) by P residual is generated by p prime elements and idempotent, this is a contradiction.

Facts & Assumptions

Given: A finite group G, a prime p, a Sylow p-subgroup P≤G controlling fusion in P with respect to G, and K:=Op(G), Q:=P∩K≠{1}.

[F1]

K⊴G, G/K is a finite p-group, KP=G, and Q=P∩K is a Sylow p-subgroup of K; in particular K is finite and [K:Q]=∣K∣/∣Q∣ is prime to p, since ∣Q∣ is the exact power of p dividing ∣K∣ (P residual of a finite group, Sylow subgroups of a normal subgroup are intersections with Sylow subgroups, Sylow p-subgroups of a finite group, If [G:N] is finite then ∣G/N∣=[G:N]; for finite G this equals ∣G∣/∣N∣, Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G).

[F2]

Q⊴P: for u∈P one has uQu−1=u(P∩K)u−1=uPu−1∩uKu−1=P∩K=Q, because uPu−1=P and K⊴G (Normal subgroup: invariance under conjugation, Subgroup, Conjugation x↦gxg−1 is an automorphism).

[F3]

Commutators: [a,b]=aba−1b−1, [A,B]=⟨[a,b]:a∈A,b∈B⟩, and [P,Q]=[Q,P] as subgroups because [p,q]=[q,p]−1; if A≤B and D≤C with A,B≤C then [A,D]≤[B,D] (Subgroup commutators and the lower central series, Commutators [g,h]=ghg−1h−1 and the commutator subgroup [G,G], The subgroup ⟨S⟩ generated by a subset, the cyclic subgroup ⟨g⟩, and cyclic groups, Subgroup, In a group e−1=e, (g−1)−1=g and (gh)−1=h−1g−1, the order of the last product being essential).

[F4]

Both P and Q are finite p-groups, Q≠{1} and Q⊴P; so by Normal p subgroup has proper commutator in a p group the commutator [Q,P] satisfies [Q,P]<Q and [Q,P]⊴Q (A finite p-group has order pn for a prime p and some n∈N, Every subgroup of a finite p-group has order a power of p).

[F7]

For q∈Q and u∈P one has quq−1=[u,q]∈[Q,P] and therefore φ(qu)=φ(q) (The conjugacy class Cl⁡G(x) and centralizer CG(x) of an element, Commutators [g,h]=ghg−1h−1 and the commutator subgroup [G,G], [F3], [F6]).

[F8]

The transfer of the homomorphism φ ⁣:Q→A is a group homomorphism V:=Vφ ⁣:K→A, independent of the transversal (Transfer homomorphism for a finite index subgroup, Transfer is a homomorphism, Transfer is independent of the transversal), and it is computed by the cycle decomposition of Transfer cycle decomposition formula: for x∈K the right cosets Q\K split into orbits C1,…,Cr of right multiplication by ⟨x⟩ of lengths ni, with representatives ti, and V(x)=∏i=1rφ(tixniti−1), where tixniti−1∈Q. The orbits partition the finite set Q\K, which has [K:Q] elements, so n1+⋯+nr=[K:Q] (The coset set G/H and the index [G:H] of a subgroup, Left and right cosets gH and Hg of a subgroup, Left group actions, transitive actions, and faithful actions, The orbits of a group action are the equivalence classes of x∼y iff y=g⋅x for some g, and hence partition the acted-on set, The orbit G⋅x and stabilizer Gx of a point in a group action).

[F10]

Every homomorphism from K to a finite p-group has Op(K) in its kernel, and Op(K)=K when K=Op(G) (P residual of a finite group, P residual is generated by p prime elements and idempotent).

Proof

technique · direct
1.1

Q is a nontrivial normal subgroup of the finite p-group P by [F1], [F2] and [F5]; so, with the commutator subgroup [Q,P] as in [F3], Normal p subgroup has proper commutator in a p group applies and gives [Q,P]<Q together with [Q,P]⊴Q. Hence A:=Q/[Q,P] is a nontrivial finite abelian p-group, and the quotient map φ ⁣:Q→A is a surjective homomorphism with ker⁡φ=[Q,P].

F3F4F5F6
2.1

Let V ⁣:K→A be the transfer of φ; by [F8] it is a group homomorphism, and for x∈K and each orbit Ci of the cycle decomposition the factor tixniti−1 lies in Q.

F8step 1.1
3.1

Fix x∈Q. For each i, the element tixniti−1=(xni)ti with ti∈K is K-conjugate to xni∈Q; both lie in Q⊆P, so the fusion-control hypothesis provides ui∈P with tixniti−1=(xni)ui. By [F7] and [F6], φ(tixniti−1)=φ(xni)=φ(x)ni.

F7F8givenstep 2.1
4.1

Consequently V(x)=∏i=1rφ(x)ni=φ(x)n1+⋯+nr=φ(x)[K:Q], the middle step because A is abelian and the last by [F8].

F5F8step 3.1
5.1

Since [Q,P]<Q, choose x∈Q∖[Q,P]; then φ(x)≠1 by [F6]. Put m:=[K:Q], which is prime to p by [F1]. If φ(x)m=1, then ord⁡(φ(x)) divides both m and ∣A∣, which is a power of p, so by [F9] ord⁡(φ(x))=1, that is φ(x)=1, a contradiction. Hence V(x)=φ(x)m≠1, and V is not the trivial homomorphism. Moreover a↦am is an endomorphism of the finite abelian group A with trivial kernel by the same order argument, so it is bijective. Since φ:Q→A is onto, step 4.1 gives V(Q)=A and therefore V:K→A is onto.

F1F6F9step 4.1
6.1

On the other hand V maps K to the finite p-group A, so Op(K)≤ker⁡V by [F10]; since K=Op(G), [F10] also gives Op(K)=K, hence K≤ker⁡V and V is trivial, contradicting step 5.1. Therefore the assumption Q≠{1} is false: Q=P∩K=P∩Op(G)={1}. ∎

F10step 5.1
TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-09-27Open item page →

Frobenius normal p complement theorem

Statement

Let G be a finite group, p a prime and P∈Syl⁡p(G) a Sylow p-subgroup (Sylow p-subgroups of a finite group). The following are equivalent.

(a) G has a normal p-complement (Normal p complement and p nilpotent group); (b) every nontrivial p-local normalizer NG(Q) with 1≠Q≤P has a normal p-complement (P local normalizer for normal complement theory); (c) P controls fusion in P with respect to G (Control of fusion in a sylow p subgroup).

Facts & Assumptions

Given: A finite group G, a prime p and a Sylow p-subgroup P≤G.

[F1]

(a) implies (b): if G has a normal p-complement, then NG(Q) has a normal p-complement for every nontrivial p-subgroup Q≤G (Normal p complements pass to subgroups and p local normalizers, P local normalizer for normal complement theory).

[F2]

(b) implies (c): if every nontrivial p-local normalizer NG(Q), 1≠Q≤P, has a normal p-complement, then P controls fusion in P with respect to G (Local normal p complements force control of fusion, Control of fusion in a sylow p subgroup).

[F3]

(c) implies (a): if P controls fusion in P with respect to G, then P∩Op(G)={1} (Fusion control forces trivial Sylow intersection with the p residual, P residual of a finite group).

[F5]

If K⊴G then PK is a subgroup of G with [G:K]=[PK:K]⋅[G:PK] in the sense that ∣G∣=∣PK∣⋅[G:PK], and PK/K≅P/(P∩K); in particular, if PK=G and P∩K={1}, then ∣G∣=∣P∣ ∣K∣ and [G:K]=∣P∣ (Second isomorphism theorem for groups: H/(H∩N)≅HN/N, First isomorphism theorem for groups: G/ker⁡f≅im⁡f, If H≤G and N⊴G, then HN is a subgroup and H∩N⊴H, If [G:N] is finite then ∣G/N∣=[G:N]; for finite G this equals ∣G∣/∣N∣, Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G, Normal subgroup: invariance under conjugation, Subgroup).

[F6]

Order facts: ∣P∣ is the exact power of p dividing ∣G∣, all Sylow p-subgroups of G have this order, and 1=p0; if S≤G then ∣S∣ divides ∣G∣ (Sylow p-subgroups of a finite group, Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G, A finite p-group has order pn for a prime p and some n∈N).

Proof

technique · direct
1.1

(a) implies (b): this is [F1].

F1
1.2

(b) implies (c): this is [F2].

F2
1.3

(c) implies (a). Assume (c). If P={1}, then ∣P∣=1=p0 is the exact power of p dividing ∣G∣ by [F6], so p∤∣G∣; then K:=G is normal in G, p∤∣K∣ and [G:K]=1=p0 is a power of p, so G has a normal p-complement.

F6given
2.1

It remains to treat the case P≠{1} under assumption (c). By [F3] we have P∩K={1} for K:=Op(G), and by [F4] K⊴G and G=PK.

F3F4step 1.3
3.1

By [F5] applied to the normal subgroup K and the subgroup P, the equality G=PK together with P∩K={1} gives ∣G∣=∣P∣ ∣K∣ and [G:K]=∣P∣.

F5step 2.1
4.1

Hence ∣K∣=∣G∣/∣P∣ is prime to p, because ∣P∣ is the exact power of p dividing ∣G∣ by [F6]; and [G:K]=∣P∣ is a power of p. Since K⊴G, the subgroup K is a normal p-complement of G.

F4F6step 2.1step 3.1
5.1

We have proved (a)⇒(b) in step 1.1, (b)⇒(c) in step 1.2, and (c)⇒(a) in steps 1.3 and 4.1; hence the three conditions are equivalent. ∎

step 1.1step 1.2step 1.3step 4.1
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

Sylow times normal subgroup covers when the index is a p-power

Statement

Let N be a finite group, p a prime, C⊴N a normal subgroup such that N/C is a p-group (Normal subgroup: invariance under conjugation, The quotient group G/N and coset product (gN)(hN)=ghN, A finite p-group has order pn for a prime p and some n∈N), and let T∈Syl⁡p(N) be a Sylow p-subgroup of N. Then N=TC.

Facts & Assumptions

Given: A finite group N, a prime p, a normal subgroup C⊴N with N/C a p-group, and a Sylow p-subgroup T≤N.

[F3]

TC is a subgroup of N with ∣TC∣=∣T∣ ∣C∣/∣T∩C∣ and TC/C≤N/C; in particular ∣TC/C∣=∣T∣/∣T∩C∣ (If H≤G and N⊴G, then HN is a subgroup and H∩N⊴H, Second isomorphism theorem for groups: H/(H∩N)≅HN/N, If [G:N] is finite then ∣G/N∣=[G:N]; for finite G this equals ∣G∣/∣N∣).

Proof

technique · direct
1.1

By [F2] the order ∣T∩C∣ is a p-power dividing ∣C∣=pa−km; since m is prime to p, ∣T∩C∣ divides pa−k, so ∣T∩C∣≤pa−k.

F1F2algebra
2.1

Hence ∣TC/C∣=∣T∣/∣T∩C∣≥pa/pa−k=pk=∣N/C∣ by [F1] and [F3].

F1F3step 1.1
3.1

Since TC/C≤N/C by [F3], the inequality of step 2.1 forces TC/C=N/C, and then ∣TC∣=∣N∣ because ∣TC∣=∣TC/C∣⋅∣C∣ by [F1] and [F3]; a subgroup of N with as many elements as N equals N, so N=TC. ∎

F1F3step 2.1
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-6-sol)audited 2026-09-27Open item page →

The local automizer condition gives centralizer conjugacy of Sylow subgroups

Statement

Let H be a finite group, p a prime, and S≤H a nontrivial p-subgroup. Put N=NH(S) and C=CH(S) (The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup, The centralizer CG(H) of a subgroup). If N/C is a p-group, then any two Sylow p-subgroups of N are conjugate by an element of C. In particular, for every x∈S the centralizer CN(x) acts transitively on the Sylow p-subgroups of N containing x.

Facts & Assumptions

Given: A finite group H, a prime p, a nontrivial p-subgroup S≤H, and the hypothesis that NH(S)/CH(S) is a p-group.

[F2]

If C⊴N and N/C is a p-group, then N=TC=CT for every T∈Syl⁡p(N) (Sylow times normal subgroup covers when the index is a p-power).

Proof

technique · direct
1.1

Let T1,T2∈Syl⁡p(N). By [F3] there is n∈N with T2=nT1n−1. Since C⊴N by [F1] and N/C is a p-group by hypothesis, [F2] gives N=CT1; write n=ct with c∈C and t∈T1.

F1F2F3given
2.1

Then T2=(ct)T1(ct)−1=c(tT1t−1)c−1=cT1c−1, since t∈T1. Thus C acts transitively on Syl⁡p(N).

F3step 1.1
3.1

Every element of C=CH(S) centralizes every x∈S, and C≤N; hence C≤CN(x) for each x∈S. The transitivity in step 2.1 therefore implies the claimed transitivity by CN(x) on the subcollection of Sylow p-subgroups containing x. ∎

F1step 2.1
LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-09-27Open item page →

P automizer condition implies fusion control

Statement

Let G be a finite group, p a prime and P∈Syl⁡p(G) a Sylow p-subgroup (Sylow p-subgroups of a finite group). Suppose that for every subgroup Q with 1≠Q≤P the quotient

NG(Q)/CG(Q)

of the normalizer by the centralizer of Q (The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup, The centralizer CG(H) of a subgroup, The quotient group G/N and coset product (gN)(hN)=ghN) is a p-group (A finite p-group has order pn for a prime p and some n∈N), where CG(Q)⊴NG(Q) by The centralizer of a normal subgroup is normal. Then P controls fusion in P with respect to G (Control of fusion in a sylow p subgroup): whenever x,y∈P and y=xg=gxg−1 for some g∈G, there is v∈P with y=xv.

Facts & Assumptions

Given: A finite group G, a prime p, a Sylow p-subgroup P≤G, and the hypothesis that NG(Q)/CG(Q) is a p-group for every subgroup Q with 1≠Q≤P.

[F1]

The hypothesis is conjugation invariant: for a subgroup R≤G and g∈G one has NG(Rg)=NG(R)g and CG(Rg)=CG(R)g, and conjugation by g induces an isomorphism NG(R)/CG(R)→NG(Rg)/CG(Rg). Since every nontrivial p-subgroup of G is conjugate into P (Sylow II: in a finite group every p-subgroup lies in a conjugate of any Sylow p-subgroup, and the Sylow p-subgroups form a single conjugacy class), the hypothesis therefore holds for every nontrivial p-subgroup R≤G (Conjugation x↦gxg−1 is an automorphism, First isomorphism theorem for groups: G/ker⁡f≅im⁡f, Group isomorphisms, automorphisms and the set Aut⁡(G), The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup, The centralizer CG(H) of a subgroup).

[F2]

If R≤G is a subgroup then CG(R)⊴NG(R), so NG(R)/CG(R) is a quotient group of NG(R) (The centralizer of a normal subgroup is normal, Normal subgroup: invariance under conjugation, The quotient group G/N and coset product (gN)(hN)=ghN).

[F3]

If S is a nontrivial p-subgroup of a finite group H and NH(S)/CH(S) is a p-group, then CH(S) acts transitively on the Sylow p-subgroups of NH(S); in particular, for each x∈S, CNH(S)(x) is transitive on those Sylow subgroups containing x (The local automizer condition gives centralizer conjugacy of Sylow subgroups).

[F4]

Local Sylow conjugacy ascent: if for every nontrivial p-subgroup S≤P and every x∈S∖{e} the centralizer CNG(S)(x) acts transitively on the Sylow p-subgroups of NG(S) containing x, then NG(P) controls fusion in P with respect to G (Local sylow conjugacy ascent for fusion, Control of fusion in a sylow p subgroup).

[F5]

If C⊴N and N/C is a p-group and T∈Syl⁡p(N), then N=TC (Sylow times normal subgroup covers when the index is a p-power, Sylow p-subgroups of a finite group).

[F8]

Conjugation laws and subgroups: xg=gxg−1, (xa)b=xba, equivalently xab=(xb)a; CG(x), CG(R) and all normalizers are subgroups, and c∈CG(R) centralizes every element of R (Conjugation x↦gxg−1 is an automorphism, In a group e−1=e, (g−1)−1=g and (gh)−1=h−1g−1, the order of the last product being essential, The conjugacy class Cl⁡G(x) and centralizer CG(x) of an element, CG(x) and NG(H) are subgroups of G, Subgroup).

Proof

technique · direct
1.1

The hypothesis holds for every nontrivial p-subgroup of G: if R≤G is a nontrivial p-subgroup, [F1] provides g∈G with Rg≤P, so NG(Rg)/CG(Rg) is a p-group and, by [F1], NG(R)/CG(R) is isomorphic to it, hence is a p-group.

F1F6
1.2

If P={1} then the only element of P is e=ee, so P controls fusion in P trivially. Assume now P≠{1}; then the hypothesis applies to the nontrivial subgroup Q:=P≤P, so NG(P)/CG(P) is a p-group, while CG(P)⊴NG(P) by [F2] and P∈Syl⁡p(NG(P)) by [F7]; hence [F5] applies with N=NG(P), C=CG(P) and the Sylow p-subgroup P, giving NG(P)=P CG(P).

F2F5F7given
1.3

Let S≤P be a nontrivial p-subgroup, put N:=NG(S) and C:=CG(S), and let x∈S∖{e}. The hypothesis directly gives that N/C is a p-group, so [F3] applies to the subgroup S≤G and gives that CN(x) acts transitively on the Sylow p-subgroups of N containing x.

F2F3given
2.1

Since S≤P and x∈S∖{e} were arbitrary, step 1.3 verifies the local centralizer-transitivity hypothesis of [F4]. Thus NG(P) controls fusion in P with respect to G.

F4step 1.3
3.1

Equivalently, for every pair a,b∈P with b=ag for some g∈G, step 2.1 supplies an element u∈NG(P) such that b=au.

F4step 2.1
4.1

Finally let x,y∈P and g∈G with y=xg. By step 3.1 there is u∈NG(P) with y=xu; by step 1.2 write u=vc with v∈P and c∈CG(P). Then, by the conjugation law of [F8], y=xvc=(xc)v=xv, because c centralizes x by [F8] and v∈P. Hence P controls fusion in P with respect to G. In this last step the hypothesis at Q=P, not only at the smaller subgroups, is what makes the conjugation action of NG(P) on P inner through the decomposition NG(P)=P CG(P) of step 1.2; for P={1} the statement is vacuous (Control of fusion in a sylow p subgroup). ∎

F8step 3.1step 1.2
CorollaryStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-09-27Open item page →

Frobenius automizer criterion for p nilpotence

Statement

Let G be a finite group, p a prime and P∈Syl⁡p(G) a Sylow p-subgroup (Sylow p-subgroups of a finite group). Then the following are equivalent.

(i) G has a normal p-complement (Normal p complement and p nilpotent group). (ii) For every subgroup Q with 1≠Q≤P, the automizer NG(Q)/CG(Q) is a p-group (The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup, The centralizer CG(H) of a subgroup, The quotient group G/N and coset product (gN)(hN)=ghN, A finite p-group has order pn for a prime p and some n∈N).

Facts & Assumptions

Given: A finite group G, a prime p and a Sylow p-subgroup P≤G.

[F1]

(ii) implies (i): if the automizer condition (ii) holds, then by P automizer condition implies fusion control the Sylow P controls fusion in P with respect to G, and then by Frobenius normal p complement theorem G has a normal p-complement (Control of fusion in a sylow p subgroup).

[F2]

(i) implies (ii): suppose G has a normal p-complement K, let Q be a subgroup with 1≠Q≤P, and put N:=NG(Q), C:=CG(Q) and K0:=K∩N. Then K⊴G with p∤∣K∣ and [G:K] a power of p, K∩P={1}, K0⊴N, Q⊴N and Q≤N (Normal p complement and p nilpotent group, The normalizer NG(H)={g∈G:gHg−1=H} of a subgroup, Normal subgroup: invariance under conjugation, Subgroup, Sylow p-subgroups of a finite group).

[F3]

Commutator inclusions used in step 1.2: if A⊴N and B≤N, then [A,B]≤A, since for a∈A, b∈B one has bab−1∈A and hence [a,b]=a(bab−1)−1∈A; and if B⊴N then [A,B]≤B likewise, since aba−1∈B; here [A,B]=⟨[a,b]:a∈A,b∈B⟩ with [a,b]=aba−1b−1 (Commutators [g,h]=ghg−1h−1 and the commutator subgroup [G,G], Subgroup commutators and the lower central series, Normal subgroup: invariance under conjugation, In a group e−1=e, (g−1)−1=g and (gh)−1=h−1g−1, the order of the last product being essential, The subgroup ⟨S⟩ generated by a subset, the cyclic subgroup ⟨g⟩, and cyclic groups).

[F5]

If P={1} then ∣P∣=1=p0 is the exact power of p dividing ∣G∣, so p∤∣G∣ and G itself is a normal p-complement of G; the condition (ii) is then vacuous (Sylow p-subgroups of a finite group, A finite p-group has order pn for a prime p and some n∈N, Lagrange's theorem: ∣G∣=[G:H]∣H∣ for every subgroup H of a finite group G, Normal p complement and p nilpotent group).

Proof

technique · direct
1.1

(ii) implies (i): this is [F1].

F1
1.2

(i) implies (ii). Assume that G has a normal p-complement K, retain the notation N=NG(Q), C=CG(Q), K0=K∩N of [F2] for a subgroup Q with 1≠Q≤P, and note K0⊴N with K0≤N and Q⊴N. By [F3] applied inside N to the normal subgroup K0 and the subgroup Q, we get [K0,Q]≤K0; applying it to the normal subgroup Q and the subgroup K0 gives [K0,Q]≤Q. Hence [K0,Q]≤K0∩Q≤K∩P={1}, so every generator of [K0,Q] is trivial and [K0,Q]={1}; that is, every element of K0 commutes with every element of Q, so K0≤C=CG(Q).

F2F3
2.1

Consequently K0≤C≤N with K0 and C normal in N: C⊴N because Q⊴N and by The centralizer of a normal subgroup is normal. By [F4] the quotient N/C is isomorphic to (N/K0)/(C/K0), a quotient of N/K0, and N/K0 is isomorphic to a subgroup of G/K.

F2F4step 1.2
3.1

Now G/K is a p-group by [F2], so its subgroup N/K0 is a p-group, and the quotient N/C of that p-group is a p-group by [F4]. As Q with 1≠Q≤P was arbitrary, (ii) holds.

F2F4step 2.1
4.1

If P={1} then both conditions hold by [F5]. Otherwise step 1.1 gives (ii)⇒(i) and step 3.1 gives (i)⇒(ii), so the two conditions are equivalent. ∎

F5step 1.1step 3.1

5 · Examples, counterexamples and false statements

None yet.

Sources