Documentation

TauCeti.MeasureTheory.Group.LocalFieldScaling

Scaling of Haar measure on ι → K under integral matrices #

Let K be a nonarchimedean local field with residue field of cardinality q, and let A be a square matrix over 𝒪[K] whose determinant is associated to ϖ ^ k for a uniformizer ϖ. Acting on ι → K by mulVec, A multiplies Haar measure by q ^ (-k). This is the nonarchimedean analogue of Mathlib's MeasureTheory.Measure.addHaar_image_linearMap, where the factor is |det| for a real vector space.

The proof runs through the lattice index. The image under A of the integer box ∏ i, 𝒪[K] is a subgroup of index q ^ k in it (Matrix.card_quotient_range_mulVecLin_eq_pow), so the box is the disjoint union of q ^ k translates of that image, and Mathlib's MeasureTheory.AddSubgroup.index_mul_measure turns this into the measure identity. On a translate of a scaled box — that is, on a ball of ι → K — the same identity holds for every left-invariant measure, with no regularity hypothesis. For a general set the measure s ↦ μ (A '' s) is again a Haar measure, so by uniqueness of Haar measure it is a constant multiple of μ, and the box computes the constant; Haar measure on ι → K is automatically regular, since ι → K is second countable and locally compact.

Specializing to diagonal matrices gives the volume of a box of ideals: the box whose i-th factor is 𝓂[K] ^ e i is the image of the integer box under the diagonal matrix with entries ϖ ^ e i, whose determinant is ϖ ^ ∑ i, e i, so it has q ^ (-∑ i, e i) times the measure of the box. The topological box lemmas apply to valuative fields without local compactness; compactness of the integer box only requires compactness of 𝒪[K].

These volumes are the change of variables in Serre's proof of his mass formula for totally ramified extensions, where A is the matrix of multiplication by an element of the ring of integers of an extension of K in an integral basis and k is the valuation of its norm.

Main definitions #

Main results #

References #

noncomputable def TauCeti.integerBox (K : Type u_1) [Field K] [ValuativeRel K] (ι : Type u_2) :
Set (ι → K)

The integer box ∏ i, 𝒪[K] of ι → K, on which Haar measure on ι → K is normalized.

Instances For
    @[simp]
    theorem TauCeti.mem_integerBox {K : Type u_1} [Field K] [ValuativeRel K] {ι : Type u_2} {a : ι → K} :
    a ∈ integerBox K ι ↔ ∀ (i : ι), a i ∈ (ValuativeRel.valuation K).integer

    Membership in the integer box: every coordinate is integral.

    theorem TauCeti.integerBox_def {K : Type u_1} [Field K] [ValuativeRel K] {ι : Type u_2} :

    The integer box is the product of copies of 𝒪[K].

    theorem TauCeti.univ_pi_image_maximalIdeal_pow_subset {K : Type u_1} [Field K] [ValuativeRel K] {ι : Type u_2} {e e' : ι → ℕ} (h : e' ≤ e) :

    Boxes of ideal powers shrink as the exponents grow.

    noncomputable def TauCeti.scaledIntegerBox {K : Type u_1} [Field K] [ValuativeRel K] (ι : Type u_2) (ϖ : K) (k : ℕ) :
    AddSubgroup (ι → K)

    The scaled integer box {x | ϖ ^ k • x ∈ integerBox K ι}, an additive subgroup of ι → K.

    Instances For
      @[simp]
      theorem TauCeti.mem_scaledIntegerBox {K : Type u_1} [Field K] [ValuativeRel K] {ι : Type u_2} {ϖ : K} {k : ℕ} {x : ι → K} :
      x ∈ scaledIntegerBox ι ϖ k ↔ ∀ (i : ι), ϖ ^ k * x i ∈ (ValuativeRel.valuation K).integer

      Membership in a scaled integer box, coordinate by coordinate.

      theorem TauCeti.coe_scaledIntegerBox {K : Type u_1} [Field K] [ValuativeRel K] {ι : Type u_2} {ϖ : K} {k : ℕ} :
      ↑(scaledIntegerBox ι ϖ k) = (fun (x : ι → K) => ϖ ^ k • x) ⁻¹' integerBox K ι

      A scaled integer box is the preimage of the integer box under multiplication by ϖ ^ k.

      theorem TauCeti.exists_mem_scaledIntegerBox {K : Type u_1} [Field K] [ValuativeRel K] [IsDiscreteValuationRing ↥(ValuativeRel.valuation K).integer] {ι : Type u_2} [Finite ι] {ϖ : ↥(ValuativeRel.valuation K).integer} (hϖ : Irreducible ϖ) (x : ι → K) :
      ∃ (k : ℕ), x ∈ scaledIntegerBox ι (↑ϖ) k

      For a uniformizer ϖ and a finite type ι, the scaled integer boxes exhaust ι → K: a power of ϖ clears the denominators of the coordinates of x, since K is the localization of 𝒪[K] away from ϖ.

      The integer box is open.

      If 𝒪[K] is compact, its integer box is compact for any index type.

      theorem TauCeti.isCompact_scaledIntegerBox {K : Type u_1} [Field K] [ValuativeRel K] [TopologicalSpace K] {ι : Type u_2} [ContinuousMul K] [CompactSpace ↥(ValuativeRel.valuation K).integer] {ϖ : K} (hϖ : ϖ ≠ 0) (k : ℕ) :

      If 𝒪[K] is compact, a scaled integer box with nonzero scale is compact: it is the preimage of the integer box under a homeomorphism.

      The scaling law on the integer box. If the determinant of a square matrix A over 𝒪[K] is associated to ϖ ^ k for a uniformizer ϖ, then for every left-invariant measure on ι → K, the image of the integer box ∏ i, 𝒪[K] under A has q ^ (-k) times the measure of the box, where q = Nat.card 𝓀[K].

      The volume of a box of ideals: for a left-invariant measure on ι → K, the box whose i-th factor is 𝓂[K] ^ e i has q ^ (-∑ i, e i) times the volume of the integer box, where q = Nat.card 𝓀[K].

      theorem MeasureTheory.Measure.measure_integerBox_inter_coord_eq_zero {K : Type u_1} [Field K] [ValuativeRel K] [TopologicalSpace K] [IsNonarchimedeanLocalField K] {ι : Type u_2} [MeasurableSpace (ι → K)] [BorelSpace (ι → K)] [Finite ι] (μ : Measure (ι → K)) [μ.IsAddLeftInvariant] (hμ : μ (TauCeti.integerBox K ι) ≠ ⊤) (j : ι) :
      μ (TauCeti.integerBox K ι ∩ {a : ι → K | a j = 0}) = 0

      A coordinate hyperplane inside the integer box is null for a left-invariant measure whose integer box has finite measure.

      theorem MeasureTheory.Measure.measure_image_mulVec_vadd_smul_integerBox {K : Type u_1} [Field K] [ValuativeRel K] [TopologicalSpace K] [IsNonarchimedeanLocalField K] {ι : Type u_2} [Fintype ι] [DecidableEq ι] [MeasurableSpace (ι → K)] [BorelSpace (ι → K)] (μ : Measure (ι → K)) [μ.IsAddLeftInvariant] {A : Matrix ι ι ↥(ValuativeRel.valuation K).integer} {ϖ : ↥(ValuativeRel.valuation K).integer} (hϖ : Irreducible ϖ) {k : ℕ} (hk : Associated A.det (ϖ ^ k)) (x : ι → K) {c : K} (hc : c ≠ 0) :

      The scaling law on balls. If the determinant of a square matrix A over 𝒪[K] is associated to ϖ ^ k for a uniformizer ϖ, then for every left-invariant measure on ι → K, the image under A of the ball x +ᵥ c • ∏ i, 𝒪[K] (with c ≠ 0) has q ^ (-k) times the measure of the ball, where q = Nat.card 𝓀[K].

      theorem MeasureTheory.Measure.addHaar_image_mulVec {K : Type u_1} [Field K] [ValuativeRel K] [TopologicalSpace K] [IsNonarchimedeanLocalField K] {ι : Type u_2} [Fintype ι] [DecidableEq ι] [MeasurableSpace (ι → K)] [BorelSpace (ι → K)] (μ : Measure (ι → K)) [μ.IsAddHaarMeasure] {A : Matrix ι ι ↥(ValuativeRel.valuation K).integer} {ϖ : ↥(ValuativeRel.valuation K).integer} (hϖ : Irreducible ϖ) {k : ℕ} (hk : Associated A.det (ϖ ^ k)) (s : Set (ι → K)) :

      The scaling law for Haar measure. If the determinant of a square matrix A over 𝒪[K] is associated to ϖ ^ k for a uniformizer ϖ, then for every additive Haar measure μ on ι → K and every set s, μ (A '' s) = q ^ (-k) * μ s, where q = Nat.card 𝓀[K]. This is the nonarchimedean analogue of MeasureTheory.Measure.addHaar_image_linearMap.