Documentation

TauCeti.RingTheory.Valuation.CofinalIdeal.Restrict

Restricting a valuation to cΓ_v(I) #

Wedhorn, Adic Spaces (arXiv:1910.05934v1), §7.1.2.

Valuation.restrictToConvex restricts a valuation to an arbitrary convex subgroup of the units of its value monoid. This file specialises it to the convex subgroup that matters for Spv (A, I): the characteristic subgroup cΓ_v(I) of an ideal, from Wedhorn's Definition 7.3.

Two things have to be arranged. cΓ_v(I) lives in the value group, so it is transported onto the units of the value monoid by ConvexSubgroup.comapUnitsWithZero; and the restriction needs cΓ_v(I) to absorb every attained value ≥ 1, which holds because it contains cΓ_v.

The point-level map on Spv A that Wedhorn's retraction r_I is built from lives in TauCeti.AlgebraicGeometry.AdicSpace.RestrictToIdeal.

Main definitions #

Main results #

The characterisation lemmas — the vanishing branches, restrictToIdeal_eq_zero_iff and restrictToIdeal_le_iff — are phrased through cΓ_v(I) itself or through vanishing of the restriction, so a consumer of them never names the transport. one_le_restrictToIdeal mentions neither, being an order fact about v.restrict.

The transport is named by four declarations, each unavoidably. RestrictedValues is the transported subgroup with a zero adjoined, and the private restrictToIdeal_def is the definitional unfolding and so carries the boundedness hypothesis over it. The two comparison lemmas against an abstract bound, coe_le_restrictToIdeal_iff and restrictToIdeal_lt_coe_iff, quantify over a member of that subgroup: their whole point is to compare a restricted value against a bound arriving from the value group rather than from a ring element, and such a bound is an element of the transported subgroup. A cofinality argument needs them in that form.

References #

Provenance #

The construction is the one called restrictIdeal in AINTLIB (github.com/CBirkbeck/AINTLIB, Apache-2.0), branch dev/adic-spaces at commit 37bbdaeb9ad9e3bc9f0d660feadc2779e455a91c, project projects/AdicSpaces/, file Adic spaces/CharacteristicSubgroup.lean, built there on restrictToConvexBounded. The port was checked against that source rather than copied: AINTLIB's cGammaIdeal is already phrased on Γ₀ˣ and so needs no transport, whereas cΓ_v(I) here lives in the value group and is carried across by ConvexSubgroup.comapUnitsWithZero.

@[reducible, inline]
abbrev Valuation.RestrictedValues {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) :
Type u_2

The value monoid of the restricted valuation: cΓ_v(I), transported onto the units of the value monoid, with a zero adjoined.

Equations
Instances For
    @[instance_reducible]
    noncomputable instance Valuation.instLinearOrderedCommGroupWithZeroRestrictedValues {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) :

    RestrictedValues is a linearly ordered commutative group with zero.

    Equations
    noncomputable def Valuation.restrictToIdeal {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) :

    Wedhorn §7.1.2: the restriction v ↦ v|cΓ_v(I). Values whose unit lies in cΓ_v(I) are kept; every other value is sent to 0.

    Equations
    Instances For

      The restriction, characterised through cΓ_v(I) #

      restrictToIdeal keeps or discards a value according to membership in the transported cΓ_v(I), which is not the form a consumer holds: the introduction rules for cΓ_v(I) speak about the value group. The bridge below converts between the two, and the lemmas after it are stated so that no consumer has to unfold the definition or mention the transport.

      theorem Valuation.restrictToIdeal_apply_of_notMem {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) {a : A} (h0 : v a ≠ 0) (hm : MonoidWithZeroHom.valueGroup.mk (↑v) 1 a ⋯ h0 ∉ v.characteristicSubgroupOfIdeal I hfg) :
      (v.restrictToIdeal I hfg) a = 0

      Off cΓ_v(I), the restriction vanishes. The hypothesis is non-membership in cΓ_v(I) itself; the transport is applied internally.

      Not @[simp]: restrictToIdeal_eq_zero_iff is the simp-normal form for a vanishing restriction, and it rewrites this lemma's left-hand side, which simpNF rejects.

      @[simp]
      theorem Valuation.restrictToIdeal_apply_of_eq_zero {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) {a : A} (h0 : v a = 0) :
      (v.restrictToIdeal I hfg) a = 0

      The restriction vanishes wherever v does.

      theorem Valuation.restrictToIdeal_eq_zero_iff_of_ne {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) {a : A} (h0 : v a ≠ 0) :

      Where the restriction vanishes at a nonzero value, stated through cΓ_v(I) itself. restrictToIdeal_eq_zero_iff is the total form, and is the @[simp] one: tagging this branch too makes its left-hand side reducible by that lemma, which simpNF rejects.

      @[simp]
      theorem Valuation.restrictToIdeal_eq_zero_iff {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) (a : A) :
      (v.restrictToIdeal I hfg) a = 0 ↔ v a = 0 ∨ ∃ (h0 : v a ≠ 0), MonoidWithZeroHom.valueGroup.mk (↑v) 1 a ⋯ h0 ∉ v.characteristicSubgroupOfIdeal I hfg

      Where the restriction vanishes, totally: at the zeros of v, and where v is nonzero but its value escapes cΓ_v(I). restrictToIdeal_eq_zero_iff_of_ne is the nonzero branch, in the form consumers holding a nonvanishing hypothesis want.

      theorem Valuation.restrictToIdeal_le_iff_of_mem {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) {a b : A} (h0a : v a ≠ 0) (h0b : v b ≠ 0) (hma : MonoidWithZeroHom.valueGroup.mk (↑v) 1 a ⋯ h0a ∈ v.characteristicSubgroupOfIdeal I hfg) (hmb : MonoidWithZeroHom.valueGroup.mk (↑v) 1 b ⋯ h0b ∈ v.characteristicSubgroupOfIdeal I hfg) :
      (v.restrictToIdeal I hfg) a ≤ (v.restrictToIdeal I hfg) b ↔ v a ≤ v b

      On values kept by the restriction, the order is both preserved and reflected — stated through membership in cΓ_v(I) itself.

      Not @[simp]: restrictToIdeal_le_iff is the simp-normal form for a comparison of restricted values, and it rewrites this lemma's left-hand side, which simpNF rejects.

      @[simp]
      theorem Valuation.restrictToIdeal_le_iff {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) (a b : A) :
      (v.restrictToIdeal I hfg) a ≤ (v.restrictToIdeal I hfg) b ↔ (v.restrictToIdeal I hfg) a = 0 ∨ (v.restrictToIdeal I hfg) b ≠ 0 ∧ v a ≤ v b

      Comparison after restriction, totally: a value discarded by the restriction is below everything, a kept value is below only kept values, and two kept values compare exactly as they did under v. restrictToIdeal_le_iff_of_mem is the both-kept branch, in the form consumers holding membership hypotheses want.

      The side conditions are stated as vanishing of the restriction, which restrictToIdeal_eq_zero_iff turns into membership in cΓ_v(I); so a caller never has to name the transported subgroup.

      theorem Valuation.coe_le_restrictToIdeal_iff {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) (a : A) (u : ↥(v.characteristicSubgroupOfIdeal I hfg).comapUnitsWithZero.toSubgroup) :
      ↑u ≤ (v.restrictToIdeal I hfg) a ↔ ↑↑u ≤ v.restrict a

      The companion of restrictToIdeal_lt_coe_iff with the member on the left.

      theorem Valuation.restrictToIdeal_lt_coe_iff {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) (a : A) (u : ↥(v.characteristicSubgroupOfIdeal I hfg).comapUnitsWithZero.toSubgroup) :
      (v.restrictToIdeal I hfg) a < ↑u ↔ v.restrict a < ↑↑u

      Comparing a restricted value against an abstract member of the transported cΓ_v(I), back in the value monoid of v. This is the form a cofinality argument needs, where the bound comes from the value group rather than from a ring element.

      theorem Valuation.one_le_restrictToIdeal {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) {a : A} (h1 : 1 ≤ v.restrict a) :
      1 ≤ (v.restrictToIdeal I hfg) a

      A value at least 1 stays at least 1: cΓ_v(I) keeps every attained value ≥ 1.

      Wedhorn §7.1.2: the restricted valuation has full characteristic subgroup for I. This is the valuation-level content of the landing law — the restriction v|cΓ_v(I) satisfies the very condition that cuts out Spv (A, I). The point-level statement is TauCeti.ValuationSpectrum.restrictToIdeal_mem_spvOfIdeal.

      The proof is Wedhorn's case split on whether I meets cΓ_v, each branch above.

      theorem Valuation.restrictToIdeal_ne_zero_of_mem {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) {a : A} (h0 : v a ≠ 0) (hmem : MonoidWithZeroHom.valueGroup.mk (↑v) 1 a ⋯ h0 ∈ v.characteristicSubgroupOfIdeal I hfg) :
      (v.restrictToIdeal I hfg) a ≠ 0

      A value whose class lies in cΓ_v(I) is kept by the restriction. The contrapositive of the nonzero branch of restrictToIdeal_eq_zero_iff, in the form a consumer holding a membership proof wants.

      theorem Valuation.restrictToIdeal_ne_zero_of_le {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) {a b : A} (hb : (v.restrictToIdeal I hfg) b ≠ 0) (hab : v b ≤ v a) :
      (v.restrictToIdeal I hfg) a ≠ 0

      The restriction keeps every value above a kept value. A value discarded by v|cΓ_v(I) cannot dominate a kept one: the kept value's class lies in cΓ_v(I), and a larger class is either below 1, so convexity puts it in, or above 1, so it is an attained characteristic value and cΓ_v ≤ cΓ_v(I) puts it in.

      This is the monotonicity that makes r_I a horizontal specialisation; Wedhorn uses it without comment in the proof of Lemma 7.5(iii).

      theorem Valuation.exists_mem_restrictToIdeal_ne_zero {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) (hne : ∃ a ∈ I, v a ≠ 0) :
      ∃ a ∈ I, (v.restrictToIdeal I hfg) a ≠ 0

      Wedhorn Lemma 7.5(3). If v does not vanish identically on I, neither does its restriction: v(I) ≠ 0 implies r_I(v)(I) ≠ 0, so I is not contained in the support of the restricted valuation. This is the form Wedhorn's Theorem 7.10 consumes.

      theorem Valuation.restrictToIdeal_ne_zero_of_isAdmissible {A : Type u_1} [CommRing A] {Γ₀ : Type u_2} [LinearOrderedCommGroupWithZero Γ₀] (v : Valuation A Γ₀) (I : Ideal A) (hfg : ∃ (J : Ideal A), J.FG ∧ I.radical = J.radical) {T : Set A} {u : A} (hadm : I ≤ (Ideal.span (insert u T)).radical) (hu0 : v u ≠ 0) (hT : ∀ t ∈ T, v t ≤ v u) :
      (v.restrictToIdeal I hfg) u ≠ 0

      The restriction does not kill an admissible denominator. If u dominates a set T with I ⊆ √((T ∪ {u}) · A) and v u ≠ 0, then v|cΓ_v(I) keeps u.

      This is one of the auxiliary claims in Wedhorn's proof of Lemma 7.5(1) — step (iii) there — and it is what lets restrictToIdealCodRestrict_preimage compute the preimage of a rational subset.