Documentation

Hironaka.Algebra.Local.Taylor

Taylor expansion and Krull's intersection theorem, in the form needed #

Kollár's proof of [Kol07, Proposition 94] takes "the Taylor expansion f(x₁+b₁, …, xₙ+bₙ) = f + ∑ᵢ bᵢ ∂f/∂xᵢ + ½ ∑ᵢⱼ bᵢbⱼ ∂²f/∂xᵢ∂xⱼ + ⋯" modulo 𝔪^{s+1} and lets s → ∞ by Krull's intersection theorem (recalled in [Kol07, Definition 55]). This file provides what that proof needs (Hironaka/Algebra/Local/Prop94.lean):

Iterated partial derivatives of polynomials #

The coordinate derivations of MvPolynomial (Fin n) A commute (the polynomial analogue of MvPowerSeries.pderiv_pderiv_comm).

@[reducible, inline]
noncomputable abbrev IsLocalRing.pderivEnd {A : Type u_1} [CommRing A] {n : ℕ} (i : Fin n) :

∂ᵢ as an A-linear endomorphism of MvPolynomial (Fin n) A.

Equations
Instances For
    theorem IsLocalRing.pderivEnd_apply {A : Type u_1} [CommRing A] {n : ℕ} (i : Fin n) (P : MvPolynomial (Fin n) A) :
    theorem IsLocalRing.pderivEnd_commute {A : Type u_1} [CommRing A] {n : ℕ} (i j : Fin n) :
    theorem IsLocalRing.pderivEnd_pow_pairwise {A : Type u_1} [CommRing A] {n : ℕ} (α : Fin n →₀ ℕ) :
    noncomputable def IsLocalRing.pderivPow {A : Type u_1} [CommRing A] {n : ℕ} (α : Fin n →₀ ℕ) :

    The iterated partial derivative ∂^α = ∏ᵢ ∂ᵢ^{αᵢ} of a multi-index α, on polynomials (the higher derivatives of the Taylor expansion in the proof of [Kol07, Proposition 94]).

    Equations
    Instances For
      theorem IsLocalRing.pderivPow_zero {A : Type u_1} [CommRing A] {n : ℕ} :
      theorem IsLocalRing.pderivPow_add {A : Type u_1} [CommRing A] {n : ℕ} (α β : Fin n →₀ ℕ) :
      theorem IsLocalRing.pderivPow_single {A : Type u_1} [CommRing A] {n : ℕ} (i : Fin n) (k : ℕ) :
      theorem IsLocalRing.pderivEnd_pow_monomial {A : Type u_1} [CommRing A] {n : ℕ} (i : Fin n) (k : ℕ) (β : Fin n →₀ ℕ) (c : A) :
      theorem IsLocalRing.pderivPow_monomial {A : Type u_1} [CommRing A] {n : ℕ} (α β : Fin n →₀ ℕ) (c : A) :
      (pderivPow α) ((MvPolynomial.monomial β) c) = (MvPolynomial.monomial (β - α)) (c * ↑(∏ i : Fin n, (β i).descFactorial (α i)))

      ∂^α X^β = (∏ᵢ βᵢ!/(βᵢ-αᵢ)!) X^{β-α} (zero unless α ≤ β).

      The factorial α! = ∏ᵢ αᵢ! of a multi-index.

      Equations
      Instances For
        theorem IsLocalRing.aeval_add_monomial {A : Type u_1} [CommRing A] {n : ℕ} {S : Type u_2} [CommRing S] [Algebra A S] [Algebra ℚ S] (g b : Fin n → S) (β : Fin n →₀ ℕ) (c : A) :
        (MvPolynomial.aeval fun (i : Fin n) => g i + b i) ((MvPolynomial.monomial β) c) = ∑ α ≤ β, (↑(multiFactorial α))⁻¹ • (MvPolynomial.aeval g) ((pderivPow α) ((MvPolynomial.monomial β) c)) * ∏ i : Fin n, b i ^ α i

        The Taylor formula for a monomial: (g + b)^β = ∑_{α ≤ β} (1/α!) (∂^α X^β)(g) b^α, the binomial theorem in each variable.

        theorem IsLocalRing.taylor_aeval_add {A : Type u_1} [CommRing A] {n : ℕ} {S : Type u_2} [CommRing S] [Algebra A S] [Algebra ℚ S] (P : MvPolynomial (Fin n) A) (g b : Fin n → S) {d : Fin n →₀ ℕ} (hd : ∀ β ∈ P.support, β ≤ d) :
        (MvPolynomial.aeval fun (i : Fin n) => g i + b i) P = ∑ α ≤ d, (↑(multiFactorial α))⁻¹ • (MvPolynomial.aeval g) ((pderivPow α) P) * ∏ i : Fin n, b i ^ α i

        The multivariate Taylor formula (Kollár's "take the Taylor expansion" in the proof of [Kol07, Proposition 94]; Włodarczyk's "the Taylor formula for n unknowns" in the proof of [Wlo05, Lemma 2.9.6]): for a polynomial P over a ℚ-algebra and g, b in a ℚ-algebra S, P(g + b) = ∑_α (1/α!) (∂^α P)(g) b^α, the sum over the multi-indices α ≤ d for any bound d on the exponents of P.

        theorem IsLocalRing.taylor_eq_sum {A : Type u_1} [CommRing A] {n : ℕ} [Algebra ℚ A] (P : MvPolynomial (Fin n) A) (b : Fin n → A) {d : Fin n →₀ ℕ} (hd : ∀ β ∈ P.support, β ≤ d) :
        (MvPolynomial.aeval fun (i : Fin n) => MvPolynomial.X i + MvPolynomial.C (b i)) P = ∑ α ≤ d, (↑(multiFactorial α))⁻¹ • (pderivPow α) P * MvPolynomial.C (∏ i : Fin n, b i ^ α i)

        The Taylor formula in Kollár's form: P(X + b) = ∑_α (1/α!) (∂^α P) b^α for P ∈ A[X], b ∈ Aⁿ.

        Iterated partial derivatives and 𝔪-adic estimates in K⟦X⟧ #

        @[reducible, inline]
        noncomputable abbrev IsLocalRing.pderivEnd' {K : Type u_1} [Field K] {n : ℕ} (i : Fin n) :

        ∂ᵢ = pderiv K i as a K-linear endomorphism of K⟦X⟧.

        Equations
        Instances For
          theorem IsLocalRing.pderivEnd'_apply {K : Type u_1} [Field K] {n : ℕ} (i : Fin n) (f : MvPowerSeries (Fin n) K) :
          theorem IsLocalRing.pderivEnd'_commute {K : Type u_1} [Field K] {n : ℕ} (i j : Fin n) :
          theorem IsLocalRing.pderivEnd'_pow_pairwise {K : Type u_1} [Field K] {n : ℕ} (α : Fin n →₀ ℕ) :
          noncomputable def IsLocalRing.pderivPowSeries {K : Type u_1} [Field K] {n : ℕ} (α : Fin n →₀ ℕ) :

          The iterated partial derivative ∂^α on K⟦X⟧.

          Equations
          Instances For
            theorem IsLocalRing.pderivEnd'_pow_coe {K : Type u_1} [Field K] {n : ℕ} (i : Fin n) (k : ℕ) (P : MvPolynomial (Fin n) K) :
            (pderivEnd' i ^ k) ↑P = ↑((pderivEnd i ^ k) P)
            theorem IsLocalRing.pderivPowSeries_coe {K : Type u_1} [Field K] {n : ℕ} (α : Fin n →₀ ℕ) (P : MvPolynomial (Fin n) K) :
            (pderivPowSeries α) ↑P = ↑((pderivPow α) P)

            ∂^α on K⟦X⟧ restricts to ∂^α on polynomials.

            theorem IsLocalRing.pderiv_mem_maximalIdeal_pow {K : Type u_1} [Field K] {n k : ℕ} {f : MvPowerSeries (Fin n) K} (hf : f ∈ maximalIdeal (MvPowerSeries (Fin n) K) ^ (k + 1)) (i : Fin n) :

            A coordinate partial lowers the 𝔪-adic order by at most one (Kollár's step ∂^α f ∈ 𝔪^{s+1-|α|} in the proof of [Kol07, Proposition 94]; the K⟦X⟧ form of RegularCoords.D_maximalIdeal_pow).

            theorem IsLocalRing.pderivEnd'_pow_mem_maximalIdeal_pow {K : Type u_1} [Field K] {n k : ℕ} {f : MvPowerSeries (Fin n) K} (hf : f ∈ maximalIdeal (MvPowerSeries (Fin n) K) ^ k) (i : Fin n) (b : ℕ) :

            ∂^α lowers the 𝔪-adic order by at most |α|.

            theorem IsLocalRing.prod_pow_mem_maximalIdeal_pow {K : Type u_1} [Field K] {n : ℕ} {b : Fin n → MvPowerSeries (Fin n) K} (hb : ∀ (i : Fin n), b i ∈ maximalIdeal (MvPowerSeries (Fin n) K)) (α : Fin n →₀ ℕ) :
            ∏ i : Fin n, b i ^ α i ∈ maximalIdeal (MvPowerSeries (Fin n) K) ^ Finsupp.degree α

            b^α ∈ 𝔪^{|α|} for bᵢ ∈ 𝔪.

            theorem IsLocalRing.sub_truncTotal_mem {K : Type u_1} [Field K] {n : ℕ} (f : MvPowerSeries (Fin n) K) (s : ℕ) :

            f ≡ truncTotal (s+1) f (mod 𝔪^{s+1}).

            theorem IsLocalRing.hasSubst_add {K : Type u_1} [Field K] {n : ℕ} {b : Fin n → MvPowerSeries (Fin n) K} (hb : ∀ (i : Fin n), b i ∈ maximalIdeal (MvPowerSeries (Fin n) K)) :

            Xᵢ ↦ Xᵢ + bᵢ is a substitution when bᵢ ∈ 𝔪.

            theorem IsLocalRing.substAlgHom_add_mem_maximalIdeal_pow {K : Type u_1} [Field K] {n : ℕ} {b : Fin n → MvPowerSeries (Fin n) K} (hb : ∀ (i : Fin n), b i ∈ maximalIdeal (MvPowerSeries (Fin n) K)) {k : ℕ} {f : MvPowerSeries (Fin n) K} (hf : f ∈ maximalIdeal (MvPowerSeries (Fin n) K) ^ k) :

            The substitution Xᵢ ↦ Xᵢ + bᵢ maps 𝔪^{s+1} into 𝔪^{s+1}.

            noncomputable def IsLocalRing.degreeLE (n s : ℕ) :

            The multi-indices of total degree ≤ s (the index set of Kollár's Taylor sum modulo 𝔪^{s+1}).

            Equations
            Instances For
              theorem IsLocalRing.substAlgHom_add_sub_taylor_mem {K : Type u_1} [Field K] {n : ℕ} [CharZero K] (f : MvPowerSeries (Fin n) K) {b : Fin n → MvPowerSeries (Fin n) K} (hb : ∀ (i : Fin n), b i ∈ maximalIdeal (MvPowerSeries (Fin n) K)) (s : ℕ) :
              (MvPowerSeries.substAlgHom ⋯) f - ∑ α ∈ degreeLE n s, (↑(multiFactorial α))⁻¹ • (pderivPowSeries α) f * ∏ i : Fin n, b i ^ α i ∈ maximalIdeal (MvPowerSeries (Fin n) K) ^ (s + 1)

              The Taylor formula modulo 𝔪^{s+1} (the proof of [Kol07, Proposition 94], in truncated form): for bᵢ ∈ 𝔪, f(X + b) ≡ ∑_{|α| ≤ s} (1/α!) (∂^α f) b^α (mod 𝔪^{s+1}).

              theorem IsLocalRing.substAlgHom_shift_sub_taylor_mem {K : Type u_1} [Field K] {n : ℕ} [CharZero K] (f : MvPowerSeries (Fin n) K) (i : Fin n) {b : MvPowerSeries (Fin n) K} (hb : b ∈ maximalIdeal (MvPowerSeries (Fin n) K)) (q : ℚ) (s : ℕ) :
              (MvPowerSeries.substAlgHom ⋯) f - ∑ j ∈ Finset.range (s + 1), q ^ j • ((↑j.factorial)⁻¹ • (pderivEnd' i ^ j) f * b ^ j) ∈ maximalIdeal (MvPowerSeries (Fin n) K) ^ (s + 1)

              The one-direction Taylor formula (the converse direction of the proof of [Kol07, Proposition 94]): f(X₁ + λb, X₂, …) ≡ ∑_{j ≤ s} λ^j (b^j/j!) ∂₁^j f (mod 𝔪^{s+1}) for b ∈ 𝔪, λ ∈ ℚ, along any coordinate i.

              Krull's intersection theorem in closed form #

              theorem IsLocalRing.mem_of_forall_mem_sup_pow {R : Type u_1} [CommRing R] [IsNoetherianRing R] [IsLocalRing R] {I : Ideal R} {f : R} (h : ∀ (s : ℕ), f ∈ I ⊔ maximalIdeal R ^ s) :
              f ∈ I

              Krull's intersection theorem in the form I = ⋂ₛ (I + 𝔪^s) recalled in [Kol07, Definition 55] and used ("letting s go to infinity") in the proof of [Kol07, Proposition 94]: in a Noetherian local ring, f ∈ I + 𝔪^s for all s implies f ∈ I.

              theorem IsLocalRing.iInf_sup_pow_eq {R : Type u_1} [CommRing R] [IsNoetherianRing R] [IsLocalRing R] (I : Ideal R) :
              ⨅ (s : ℕ), I ⊔ maximalIdeal R ^ s = I

              Krull's intersection theorem in closed form: I = ⋂ₛ (I + 𝔪^s).