Documentation

Hironaka.Scheme.Smooth.ExceptionalModel

The exceptional divisor of the model blow-up is smooth #

The model case of the smoothness of the exceptional divisor of a smooth blow-up: the blow-up B_L 𝔸ⁿ_A of 𝔸ⁿ_A = Spec A[x_0, …, x_{n-1}] along the coordinate subspace L = V(x_0, …, x_{r-1}), with exceptional divisor F_L = π⁻¹(L) (the blow-up via the Rees algebra, [Hau14, Definition 4.7]). Kollár's sentence "if π_{Z,X} is a smooth blow-up, then F … [is] smooth" [Kol07, Notation 19] is stated without proof; Hironaka/Scheme/Smooth/ExceptionalChart.lean and ExceptionalDivisor.lean reduce it to the model, and this file proves the model.

Why the theorems hold. In the chart U_j (j < r) the exceptional ideal pulls back along the chart U_j ≅ 𝔸ⁿ_A to the ideal (x_j), so F_L ∩ U_j = V(x_j) = Spec A[x]/(x_j) ≅ 𝔸^{n-1}_A, smooth over A of relative dimension n − 1 (smoothOfRelativeDimension_quotient_span_X); and through the chart map π_j (x_i ↦ x_i x_j for i < r, i ≠ j) the ring A[x]/(x_j) is the polynomial ring over 𝒪(L) = A[x]/I in the r − 1 variables x_i, i < r, i ≠ j (fibreAlgEquivOverCenter), so V(x_j) → L is smooth of relative dimension r − 1 (cf. Hau14, Example 5.9). The charts U_j, j < r, cover B_L 𝔸ⁿ, hence the pieces F_L ∩ U_j cover F_L, and both smoothness statements are Zariski-local on the source. Two bookkeeping identifications make this a proof about the glued blow-up: the closed subscheme of the ideal sheaf of an ideal J on Spec S is Spec (S/J) (specQuotientToSubscheme, an isomorphism because both are closed immersions with the same kernel), and the glued blow-up blowUp (coordinateSubspace k n r) is identified with the affine model over 𝔸ⁿ by the uniqueness of blow-ups (exists_iso_blowUp_coordinateSubspace), which carries the exceptional divisors and the maps to L along (prop_subschemeι_comp_iff_of_eq_comap, prop_subschemeMap_iff_of_eq_comap).

Main declarations #

The closed immersion Spec (R/I) → Spec R factored through the closed subscheme V(I) of the ideal sheaf specIdealSheaf I of I: an isomorphism, both being closed immersions with kernel the ideal sheaf of I.

Equations
  • One or more equations did not get rendered due to their size.
Instances For

    The exceptional ideal of the affine blow-up contains the pullback of the center's ideal sheaf: I·𝒪_B is the inverse image of I, so I ≤ π_*(I·𝒪_B) (Mathlib's le_map_comap, stated with the exceptional ideal's name so that subschemeMap terms carry a syntactically matching proof).

    A morphism property of V(J) → Y, for J the ideal sheaf of an ideal I on Spec R, is read on Spec (R/I) → Y.

    A morphism property of the map of closed subschemes V(J) → V(I) induced by Spec φ : Spec S → Spec R (φ : R → S carrying I into J) is read on Spec (R/I → S/J).

    The closed subscheme of the inverse image of J along an isomorphism maps isomorphically onto V(J) (Mathlib's isPullback_of_isClosedImmersion).

    A morphism property of V(K) → Y transports along an isomorphism e : B ≅ B' carrying K to J (K = J.comap e).

    theorem AlgebraicGeometry.prop_subschemeMap_iff_of_eq_comap {B B' : Scheme} (e : B ⟶ B') [CategoryTheory.IsIso e] (J : B'.IdealSheafData) (P : CategoryTheory.MorphismProperty Scheme) [P.RespectsIso] {Y : Scheme} (K : B.IdealSheafData) (hK : K = J.comap e) {q : B' ⟶ Y} {q' : B ⟶ Y} (hq : q' = CategoryTheory.CategoryStruct.comp e q) (L : Y.IdealSheafData) (H1 : L ≤ K.map q') (H2 : L ≤ J.map q) :
    P (K.subschemeMap L q' H1) ↔ P (J.subschemeMap L q H2)

    A morphism property of the induced map V(K) → V(L) transports along an isomorphism e : B ≅ B' carrying K to J.

    The piece F_L ∩ U_j of the model's exceptional divisor over the chart U_j, as the closed subscheme of the inverse image ideal, mapping into F_L (an open immersion).

    Equations
    • One or more equations did not get rendered due to their size.
    Instances For

      The pieces F_L ∩ U_j, j < r, cover F_L (the charts cover B_L 𝔸ⁿ, iSup_opensRange_modelChart).

      The open cover of the model's exceptional divisor F_L by its chart pieces F_L ∩ U_j, j < r.

      Equations
      • One or more equations did not get rendered due to their size.
      Instances For

        Over Spec A the chart U_j is Spec A[x] → Spec A (the chart map is an A-algebra map).

        The exceptional divisor F_L of B_L 𝔸ⁿ_A is smooth over A of relative dimension n − 1 (the model case of [Kol07, Notation 19]): on the piece F_L ∩ U_j = Spec A[x]/(x_j) ≅ 𝔸^{n-1}_A by the chart computation, and smoothness is Zariski-local on the source.

        theorem AlgebraicGeometry.CoordinateSubspace.card_lt_ne (n r : ℕ) (j : Fin n) (hj : ↑j < r) (hrn : r ≤ n) :
        Nat.card { i : Fin n // ↑i < r ∧ i ≠ j } = r - 1

        {i < r, i ≠ j} has r − 1 elements when j < r ≤ n.

        In the chart U_j: V(x_j) → L, Spec of the map A[x]/I → A[x]/(x_j) induced by the chart map, is smooth of relative dimension r − 1, A[x]/(x_j) being the polynomial ring over 𝒪(L) in the r − 1 variables x_i, i < r, i ≠ j (cf. [Hau14, Example 5.9]).

        On the chart U_j: the piece F_L ∩ U_j → L of F_L → L is Spec (A[x]/I → A[x]/(x_j)), smooth of relative dimension r − 1.

        F_L → L is smooth of relative dimension r − 1: on the piece F_L ∩ U_j it is Spec (A[x]/I → A[x]/(x_j)), and smoothness is Zariski-local on the source.

        Under the identification e of the glued blowUp (coordinateSubspace k n r) with the model (exists_iso_blowUp_coordinateSubspace), the exceptional divisor F_L of the glued blow-up is the inverse image of the model's exceptional ideal.

        The exceptional divisor of the glued blowUp (coordinateSubspace k n r) is smooth over k of relative dimension n − 1.

        F_L → L is smooth of relative dimension r − 1 for the glued blowUp (coordinateSubspace k n r).