Documentation

TauCeti.AlgebraicGeometry.AdicSpace.Spa.Localization.LaurentCover.Uniform

The Laurent cover of a uniform Tate ring #

Let A be a complete Hausdorff Tate ring and f ∈ A. The rational subsets

U₁ = R({f, 1}/1) = {|f| ≤ 1},      U₂ = R({1}/f) = {|f| ≥ 1},      U₁ ∩ U₂ = R({f², f, 1}/(1 · f))

cover Spa(A, A⁺). When A is uniform, that is, when its power-bounded elements form a bounded set, the augmented Čech sequence of this cover,

A → A⟨U₁⟩ × A⟨U₂⟩ → A⟨U₁ ∩ U₂⟩,      a ↦ (a, a),      (x, y) ↦ x|U₁∩U₂ - y|U₁∩U₂,

is exact, and its first map is a closed embedding. These are the injectivity and the exactness in the middle in Corollary 4 of Buzzard–Verberkmoes, the case of a Laurent cover in their theorem that stably uniform Tate rings are sheafy (their Theorem 7); Corollary 4 also asserts that the second map is surjective, which is not part of the results here. The statements have the same form as laurentCover_exact and isClosedEmbedding_laurentCover, which assume strong noetherianness instead of uniformity.

Main results #

Implementation notes #

The rings of definition of the three localisations are D₁ = A₀[f], D₂ = A₀[1/f] and D₁₂ = A₀[f, 1/f]. The first result is Buzzard–Verberkmoes's Lemma 3 for the cover {1, f}: an element of D₁ ∩ D₂ is, up to a power of f, a polynomial in f of bounded degree on both sides, so its powers all lie in one finitely generated A₀-submodule. With uniformity it bounds the elements of A lying in ϖⁿ D₁ and in ϖⁿ D₂ by ϖⁿ A°, so A carries the topology induced from A⟨U₁⟩ × A⟨U₂⟩, and the image of the complete ring A is closed. Since D₁₂ = D₁ + D₂, the difference map is strict before completion, which makes every pair agreeing on U₁ ∩ U₂ a limit of pairs coming from A. Buzzard–Verberkmoes's Lemma 2 obtains this step from the general fact that completion preserves exact sequences of strict maps; here it is proved directly for this sequence.

References #

Polynomials of bounded degree in one element #

Power-bounded elements on the two Laurent pieces #

theorem TauCeti.ValuationSpectrum.isPowerBounded_of_algebraMap_mem_locSubring_laurentCover {A : Type u_1} [CommRing A] [TopologicalSpace A] (P : Huber.PairOfDefinition A) [IsTopologicalRing A] (f : A) {S₁ : Type u_2} {S₂ : Type u_3} [CommRing S₁] [Algebra A S₁] [IsLocalization.Away 1 S₁] [CommRing S₂] [Algebra A S₂] [IsLocalization.Away f S₂] {a : A} (h₁ : (algebraMap A S₁) a ∈ P.locSubring {f, 1} 1 S₁) (h₂ : (algebraMap A S₂) a ∈ P.locSubring {1} f S₂) :

Buzzard–Verberkmoes, Lemma 3, for the Laurent cover of f. Let A₀ be the ring of definition of a pair of definition of A, and f ∈ A. If a ∈ A lies in A₀[f], the ring of definition of R({f, 1}/1) = {|f| ≤ 1}, and its image in A[1/f] lies in A₀[1/f], the ring of definition of R({1}/f) = {|f| ≥ 1}, then a is power-bounded in A. Compare isPowerBounded_of_mem_locSubring, which concerns power-boundedness in a localisation Aₛ.

The overlap ring of definition is the sum of those of the two pieces #

Approximation on the Laurent cover #

The Laurent cover of a uniform Tate ring #

Buzzard–Verberkmoes, Lemma 2 and Corollary 4, strictness. Let A be a complete Hausdorff uniform Tate ring and f ∈ A. The map A → A⟨U₁⟩ × A⟨U₂⟩, a ↦ (a, a), into the coordinate rings of U₁ = R({f, 1}/1) = {|f| ≤ 1} and U₂ = R({1}/f) = {|f| ≥ 1} is a closed embedding: it is injective, its image is closed, and the topology of A is the one induced from the product. This is isClosedEmbedding_laurentCover with uniformity in place of strong noetherianness.

theorem TauCeti.ValuationSpectrum.laurentCover_exact_of_isUniform {A : Type u_1} [CommRing A] [UniformSpace A] [IsUniformAddGroup A] [IsTopologicalRing A] [Huber.IsTateRing A] [Huber.IsUniform A] (P : Huber.PairOfDefinition A) (f : A) (S₁ : Type u_2) [CommRing S₁] [Algebra A S₁] [IsLocalization.Away 1 S₁] (S₂ : Type u_3) [CommRing S₂] [Algebra A S₂] [IsLocalization.Away f S₂] (hden₂ : P.HasDenominatorPower {1} f S₂) [CompleteSpace A] [T0Space A] (S₁₂ : Type u_4) [CommRing S₁₂] [Algebra A S₁₂] [IsLocalization.Away (1 * f) S₁₂] :
Function.Exact ⇑((P.toCompletionLoc {f, 1} 1 S₁ ⋯).prod (P.toCompletionLoc {1} f S₂ hden₂)) ⇑((P.restrictionRingHom {f, 1} 1 S₁ ⋯ {f * f, f, 1} (1 * f) S₁₂ ⋯ f ⋯ ⋯).toAddMonoidHom.comp (AddMonoidHom.fst (UniformSpace.Completion S₁) (UniformSpace.Completion S₂)) - (P.restrictionRingHom {1} f S₂ hden₂ {f * f, f, 1} (1 * f) S₁₂ ⋯ 1 ⋯ ⋯).toAddMonoidHom.comp (AddMonoidHom.snd (UniformSpace.Completion S₁) (UniformSpace.Completion S₂)))

Buzzard–Verberkmoes, Corollary 4, exactness in the middle. Let A be a complete Hausdorff uniform Tate ring and f ∈ A, and let U₁ = R({f, 1}/1), U₂ = R({1}/f) and U₁ ∩ U₂ = R({f², f, 1}/(1 · f)). In

A → A⟨U₁⟩ × A⟨U₂⟩ → A⟨U₁ ∩ U₂⟩,      a ↦ (a, a),      (x, y) ↦ x|U₁∩U₂ - y|U₁∩U₂,

the kernel of the second map is the image of the first. The restriction maps are those of the refinements with cofactors f and 1. The first map is injective, and even a closed embedding (isClosedEmbedding_laurentCover_of_isUniform). Only U₂ comes with a standing hypothesis: the one for U₁ is automatic at the denominator 1 (TauCeti.Huber.PairOfDefinition.hasDenominatorPower_denom_one), and the one for U₁ ∩ U₂ is built from those two by TauCeti.Huber.PairOfDefinition.hasDenominatorPower_mul. This is laurentCover_exact with uniformity in place of strong noetherianness.