Documentation

TauCeti.Topology.Algebra.Group.Profinite.Demushkin.QInvariant

The abelianization of a Demushkin group and Labute's q-invariant #

A Demushkin group G of rank n is a one-relator pro-p group, presented on n generators by a single relator r lying in the Frattini subgroup of the free pro-p group. Abelianizing that presentation exhibits the topological abelianization as

G^{ab} ≅ ℤ_p^{n-1} × ℤ_p ⧸ q ℤ_p

for a coordinate q of the exponent vector of r (Labute, p. 106). Because r lies in the Frattini subgroup, q is divisible by p, so the factor ℤ_p ⧸ q ℤ_p is either ℤ_p (when q = 0) or a finite cyclic group of order p ^ v_p(q) ≥ p. Consequently the torsion subgroup of G^{ab} is finite and cyclic, and it is trivial exactly when q = 0.

Labute's q-invariant demushkinQ hG is read off from the topological abelianization alone: it is 0 when G^{ab} is torsion-free and the order of the torsion subgroup otherwise. The structure theorem is then restated with this invariant as the modulus, G^{ab} ≅ ℤ_p^{n-1} × ℤ_p ⧸ (q(G)), which shows that q(G) is the invariant q of Labute's classification, an isomorphism invariant divisible by p. It is the first of the two invariants, together with the rank, that classify Demushkin groups with q ≠ 2.

Main definitions #

Main results #

References #

noncomputable def TauCeti.demushkinQ {p : ℕ} [Fact (Nat.Prime p)] {G : Type u} [Group G] [TopologicalSpace G] [IsTopologicalGroup G] (_hG : IsDemushkin p G) :

Labute's q-invariant of a Demushkin group. It is 0 when the topological abelianization G^{ab} is torsion-free, which is Labute's q = p^∞ convention, and the number of torsion elements of G^{ab} otherwise. For a Demushkin group of rank n the torsion subgroup of G^{ab} is finite and cyclic, and G^{ab} ≅ ℤ_p^{n-1} × ℤ_p ⧸ (q) (TauCeti.IsDemushkin.nonempty_continuousMulEquiv_topologicalAbelianization).

Equations
Instances For

    The q-invariant vanishes when the abelianization is torsion-free.

    When the abelianization is not torsion-free, the q-invariant is the number of its torsion elements.

    The q-invariant is an isomorphism invariant.

    The q-invariant read off a model of the abelianization #

    The model with q = 0 is torsion-free. If G^{ab} ≅ ℤ_p^ι × ℤ_p ⧸ (q) with q = 0, then G^{ab} is torsion-free.

    The q-invariant vanishes on the torsion-free model. If G^{ab} ≅ ℤ_p^ι × ℤ_p ⧸ (q) with q = 0, then q(G) = 0.

    The q-invariant is p^{v_p(q)} on the model with torsion. If G^{ab} ≅ ℤ_p^ι × ℤ_p ⧸ (q) with q ≠ 0 divisible by p, then the torsion subgroup of G^{ab} is the cyclic group ℤ_p ⧸ (q) of order p^{v_p(q)}, and that order is q(G).

    The torsion subgroup of the abelianization of a Demushkin group is finite: the abelianization is a topologically finitely generated abelian pro-p group.

    @[simp]

    The q-invariant vanishes exactly when the abelianization is torsion-free.

    The torsion subgroup of the abelianization of a Demushkin group is cyclic.

    The q-invariant of a Demushkin group is divisible by p, because the relator of a minimal presentation lies in the Frattini subgroup.

    The abelianization structure theorem for Demushkin groups (Labute, p. 106): for a Demushkin group G of rank n with q-invariant q,

    G^{ab} ≃ₜ* ℤ_p^{n-1} × ℤ_p ⧸ q ℤ_p

    as topological groups. When q = 0 the second factor is ℤ_p and G^{ab} ≅ ℤ_p^n is torsion-free; otherwise it is the cyclic group of order q, which is the torsion subgroup of G^{ab}.

    A nonzero q-invariant is a positive power of p: it is the order of the finite cyclic p-group ℤ_p ⧸ (q), the torsion subgroup of the abelianization.

    A q-invariant other than 0 and p is p ^ k with k ≥ 2.

    One-relator presentations #

    For a Demushkin group given by a one-relator presentation ⟨X ∣ r⟩, the q-invariant is read off the exponent vector exponentSum r = q • w, w x₀ = 1, of the relator, through the one-relator abelianization structure theorem TauCeti.presentedProP.oneRelatorAbelianizationEquiv: it is 0 when q = 0, that is when r lies in the closed commutator subgroup, and p^{v_p(q)} otherwise. The hypothesis p ∣ q is not decoration: the presentation need not be minimal, and a relator with a unit coordinate, such as x₃ (x₁, x₂) presenting ℤ_p × ℤ_p, has q(G) = 0 while ℤ_p ⧸ (q) is trivial.

    theorem TauCeti.demushkinQ_presentedProP_eq_zero_iff {p : ℕ} [Fact (Nat.Prime p)] {X : Type u} [Finite X] {r : freeProP p X} {x₀ : X} {w : X → ℤ_[p]} {q : ℤ_[p]} (hw : w x₀ = 1) (hr : Multiplicative.toAdd ((freeProP.exponentSum p X) r) = q • w) (hG : IsDemushkin p (presentedProP p X {r})) (hpq : ↑p ∣ q) :
    demushkinQ hG = 0 ↔ q = 0

    The q-invariant of a one-relator Demushkin group vanishes exactly when the exponent coordinate q of its relator does, for p ∣ q.

    theorem TauCeti.demushkinQ_presentedProP_eq_zero_iff_mem_topologicalClosure_commutator {p : ℕ} [Fact (Nat.Prime p)] {X : Type u} [Finite X] {r : freeProP p X} {x₀ : X} {w : X → ℤ_[p]} {q : ℤ_[p]} (hw : w x₀ = 1) (hr : Multiplicative.toAdd ((freeProP.exponentSum p X) r) = q • w) (hG : IsDemushkin p (presentedProP p X {r})) (hpq : ↑p ∣ q) :

    The q-invariant of a one-relator Demushkin group vanishes exactly when the relator lies in the closed commutator subgroup of the free pro-p group, for p ∣ q.

    theorem TauCeti.demushkinQ_presentedProP_eq_pow_valuation {p : ℕ} [Fact (Nat.Prime p)] {X : Type u} [Finite X] {r : freeProP p X} {x₀ : X} {w : X → ℤ_[p]} {q : ℤ_[p]} (hw : w x₀ = 1) (hr : Multiplicative.toAdd ((freeProP.exponentSum p X) r) = q • w) (hG : IsDemushkin p (presentedProP p X {r})) (hpq : ↑p ∣ q) (hq : q ≠ 0) :

    The q-invariant of a one-relator Demushkin group is p^{v_p(q)} for the exponent coordinate q ≠ 0 of its relator, p ∣ q: the torsion subgroup of G^{ab} is ℤ_p ⧸ (q).

    theorem TauCeti.demushkinQ_presentedProP_eq_iff_exists_not_dvd {p : ℕ} [Fact (Nat.Prime p)] {X : Type u} [Finite X] {r : freeProP p X} (hG : IsDemushkin p (presentedProP p X {r})) (hpr : ∀ (x : X), ↑p ∣ Multiplicative.toAdd ((freeProP.exponentSum p X) r) x) :
    demushkinQ hG = p ↔ ∃ (x : X), ¬↑p ^ 2 ∣ Multiplicative.toAdd ((freeProP.exponentSum p X) r) x

    The q-invariant of a one-relator Demushkin group is p exactly when some exponent sum of the relator is not divisible by p ^ 2, for a relator all of whose exponent sums are divisible by p, as they are for a relator in Φ(F). Writing the exponent vector as q • w with a coordinate w x₀ = 1, the q-invariant is p^{v_p(q)}, and it is p exactly when v_p(q) = 1.

    The q-invariant is p exactly when the relator has a p-power part. For a relator r ∈ Φ(F) presenting a Demushkin group, the q-invariant of the group is p exactly when the class of r in gr_1(F) has a nonzero p-power coordinate, that is, when some exponent sum of r is not divisible by p ^ 2.