Documentation

TauCeti.RingTheory.Huber.WeightedRestrictedSeries.Complete

Completeness of the weighted restricted power series over a complete base #

Over a nonarchimedean ring A whose weights Tν = T₁^ν₁ ⋯ Tₖ^νₖ are bounded (TauCeti.Huber.IsBounded), the weighted restricted power-series ring TauCeti.Huber.weightedRestrictedSubring is Hausdorff whenever A is, and complete over a complete uniform A. Boundedness is the whole of what the arguments need beyond Wedhorn's standing hypothesis: the neighbourhood subgroup U⟨X⟩ maps into Tν · U coefficientwise, and boundedness of Tν says exactly that Tν · U still shrinks to zero with U, so the coefficient maps are continuous — and uniformly continuous once A carries a uniformity. Points are then separated coefficientwise, and a Cauchy filter of restricted series is coefficientwise Cauchy, its coefficientwise limit is again restricted, and the filter converges to it. Both facts are Wedhorn's Proposition 5.49 (Adic Spaces, arXiv:1910.05934v1) — Hausdorffness is its part (2), and completeness is the step its part (3) is proved by, the one that then identifies A⟨X⟩_T with the completion of A[X]_T. Its part (1), density of the polynomials, is TauCeti.Huber.dense_weightedPolynomials in the parent module.

The trivial weight family Tᵢ = {1} is bounded, since Tν is then the singleton {1}, so the ordinary restricted power-series ring A⟨X⟩ gets each of the four facts by specialisation. It is that case the comparison below runs on, and the two T0Space/CompleteSpace statements are registered as instances there.

Hausdorffness is purely topological — it uses only continuity of the coefficient maps — so it is proved in a section over a nonarchimedean ring with no uniformity of its own, which is the setting the rest of the Huber development works in; the uniform hypotheses on A enter only with completeness.

Consequently TauCeti.Huber.restrictedMvPowerSeriesCompletion k A collapses: TauCeti.Huber.restrictedMvPowerSeriesCompletionEquiv identifies A⟨X₁,…,Xₖ⟩ with the plain restricted-series ring — the "comparison with the usual completed restricted power-series algebra" milestone of roadmap Layer 0.5. Its hypotheses are completeness and Hausdorffness of that ring itself rather than of A, which the instances here supply over a complete Hausdorff base and which hold over a discrete base too, so that the comparison also covers TauCeti.Huber.IsStronglyNoetherian.of_discreteTopology. It is packaged twice over one and the same underlying map — as that ring isomorphism and as the A-algebra equivalence TauCeti.Huber.restrictedMvPowerSeriesCompletionAlgEquiv, the two tied together by TauCeti.Huber.coe_restrictedMvPowerSeriesCompletionAlgEquiv and TauCeti.Huber.coe_restrictedMvPowerSeriesCompletionAlgEquiv_symm — and both it and its inverse are uniformly continuous, hence continuous by UniformContinuous.continuous. Nothing in that block is special to restricted series: every declaration in it specializes a generic statement about UniformSpace.Completion.completeRingEquivSelf proved in TauCeti.Topology.Algebra.UniformRing.

Three steps are named as private lemmas rather than exported as API. That the subring's uniformity is the one its subgroup basis induces is definitional, and is named so that the argument does not silently depend on the two being reducibly equal; it specializes Mathlib's AddGroupFilterBasis.cauchy_iff, with a single use here. The passage of a Cauchy filter to its coefficientwise limits — where Tν · U has to be closed, by TauCeti.Huber.IsWeightFamily.isOpen_weightMul and AddSubgroup.isClosed_of_isOpen — is used twice in the completeness proof. Boundedness of the trivial weight, read off TauCeti.Huber.weightPow_one_weight and TauCeti.Huber.isBounded_singleton, is what all four specialisations are discharged with.

Main results #

Provenance #

AINTLIB (github.com/CBirkbeck/AINTLIB, Apache-2.0) is the roadmap's designated prior formalisation for this row. At commit 2baa76f742bdb4fb8ee323fabba41203bd390e08 its file projects/AdicSpaces/Adic spaces/RestrictedPowerSeries.lean states nothing about completeness, Hausdorffness, or the completion of the restricted-series ring, so there was nothing to port here; nothing was copied.

References #

The coefficient maps and Hausdorffness #

Neither fact needs a uniformity on A: they hold over any nonarchimedean topological ring, which is how A is fixed throughout the rest of the Huber development.

theorem TauCeti.Huber.IsWeightFamily.continuous_coeff {k : ℕ} {A : Type u_1} [CommRing A] [TopologicalSpace A] [NonarchimedeanRing A] {T : Fin k → Set A} (hT : IsWeightFamily T) {ν : Fin k →₀ ℕ} (hb : IsBounded (weightPow T ν)) :

The ν-th coefficient map of A⟨X⟩_T is continuous as soon as the weight Tν is bounded: the neighbourhood subgroup U⟨X⟩ maps into Tν · U coefficientwise, and boundedness is exactly what makes Tν · U shrink with U.

A⟨X⟩_T over a Hausdorff base is Hausdorff (Wedhorn 5.49(2)) as soon as every weight is bounded: the coefficient maps are then continuous, and they separate points.

theorem TauCeti.Huber.continuous_coeff_one_weight {k : ℕ} {A : Type u_1} [CommRing A] [TopologicalSpace A] [NonarchimedeanRing A] (ν : Fin k →₀ ℕ) :
Continuous fun (f : ↥(weightedRestrictedSubring (fun (x : Fin k) => {1}) ⋯)) => (MvPowerSeries.coeff ν) ↑f

At the trivial weight family the coefficient maps of A⟨X⟩ are continuous: the neighbourhood subgroup U⟨X⟩ maps into U coefficientwise.

Restricted series over a Hausdorff base are Hausdorff (Wedhorn 5.49(2) at the trivial weight family), the case registered as an instance.

Completeness #

The ν-th coefficient map of A⟨X⟩_T is uniformly continuous as soon as the weight Tν is bounded: it is then a continuous additive group homomorphism.

A⟨X⟩_T is complete over a complete base when every weight is bounded. This is the step Wedhorn 5.49(3) is proved by: a Cauchy filter is coefficientwise Cauchy, its coefficientwise limit is again T-restricted, and the filter converges to it.

At the trivial weight family the coefficient maps of A⟨X⟩ are uniformly continuous: they are continuous additive group homomorphisms.

The trivial-weight restricted-series subring is complete whenever the base uniform nonarchimedean commutative ring is complete, the case registered as an instance.

The comparison with A⟨X₁,…,Xₖ⟩ #

The hypotheses are completeness and Hausdorffness of the restricted-series ring itself, not of A: the instances above supply them over a complete Hausdorff base, and over a discrete base they hold because the ring is then discrete. Each declaration specializes a generic statement about UniformSpace.Completion.completeRingEquivSelf.

The comparison equivalence of roadmap Layer 0.5: when the plain restricted-series ring is complete and Hausdorff, the completed restricted power-series algebra A⟨X₁,…,Xₖ⟩ is that ring.

Equations
Instances For
    @[simp]

    The comparison undoes the canonical inclusion: on a restricted series regarded as an element of the completion, it returns that series.

    @[simp]

    The inverse comparison is the canonical inclusion: it sends a restricted series to itself, regarded as an element of the completion.

    The inverse comparison is the canonical map into the completion, as a function.

    The comparison is uniformly continuous: it is the uniform bijection between a complete Hausdorff space and its completion.

    The inverse comparison is uniformly continuous: it is the canonical map into the completion.

    The comparison as an A-algebra equivalence: the same identification, structure map included. It is TauCeti.Huber.restrictedMvPowerSeriesCompletionEquiv rebundled, so the two share an underlying map and the coercion lemmas for the latter apply to it.

    Equations
    Instances For
      @[simp]

      The A-algebra equivalence has the same underlying map as the ring equivalence, so simp normalises the algebra bundling onto the ring one.

      @[simp]

      The inverses agree too, so the two bundlings normalise together in both directions.