Documentation

Hironaka.Scheme.BlowUp.Transform.Defs

Transforms of ideal sheaves under a blow-up #

For the blow-up π : W' → W along Z with exceptional divisor E, an ideal I on W has the total transform I* = I·𝒪_{W'} [Hau14, Definition 6.1], the strict transform Iˢ = ⋃_{i ≥ 0} (I* : I_E^i), the union of colon ideals [Hau14, Definition 6.2], and, for c at most the order of I along Z, the controlled transform I^!, the unique ideal with I* = I_E^c · I^! [Hau14, Definition 6.8], whose case c = ord_Z I is the weak transform I^g. Hironaka defines the weak transform under a monoidal transformation with non-singular irreducible centre D as the quotient of f⁻¹(J) by the m-th power of the ideal of f⁻¹(D), m = ν(J_x) at the generic point x of D, noting that m = ν(f⁻¹(J)_y) at the generic point y of f⁻¹(D) [Hir64, Ch. 0, §5, p. 142]; so m is the largest power of that ideal dividing f⁻¹(J), the exponent used here. The strict transform of a closed subscheme is the subscheme cut out by the sections supported on the exceptional divisor, i.e. the I_E-power torsion [TSPA, Tag 080D].

The transforms are defined along any morphism π : B ⟶ X with a distinguished ideal sheaf E on B, from the colon and the saturation of ideal sheaves: strictTransformAlong π E J, controlledTransformAlong π E J c, exceptionalOrderAlong π E J (the supremum in ℕ of the exponents c with E ^ c ∣ J.comap π, 0 when unbounded, i.e. when J.comap π = 0) and weakTransformAlong π E J. Along the blow-up D.blowUpπ with E = D.exceptionalDivisor they are the transforms of the main theorems: Hironaka's weak transform J.weakTransform D (with the convention for a disconnected centre stated on its docstring), the strict transform Y.strictTransform D of a closed subscheme, and Hironaka's red(f⁻¹(E) ∪ f⁻¹(D)), E.reducedTransform D.

Boundary cases: J.comap π = 0 gives exceptional order 0 and weak transform 0; c = 0 gives the total transform as controlled transform.

The strict transform of J along π with respect to the divisor E: the E-power torsion saturation of the total transform J.comap π [Hau14, Definition 6.2]; [TSPA, Tag 080D]. The strict transform of a closed subscheme of X under the blow-up is this along D.blowUpπ with E the exceptional ideal.

Equations
Instances For

    The controlled transform of J with control c, (J* : E^c) [Hau14, Definition 6.8]; [Kol07, Definition 60, (60.1)]; for E ^ c ∣ J* it is the unique ideal sheaf with J* = E^c · J^! (pow_mul_colon_of_dvd, colon_pow_eq_of_mul_eq).

    Equations
    Instances For
      noncomputable def AlgebraicGeometry.exceptionalOrderAlong {B X : Scheme} (π : B ⟶ X) (E : B.IdealSheafData) (J : X.IdealSheafData) :

      The exceptional order: the largest m such that E ^ m divides J.comap π, as a natural number (0 if unbounded, i.e. J* = 0). For a non-singular irreducible centre in a non-singular scheme this is ν(J) at the generic point of the centre [Hir64, Ch. 0, §5, p. 142].

      Equations
      Instances For

        The weak transform, the controlled transform with the exceptional order as control [Hir64, Ch. 0, §5, p. 142]; [Hau14, Definition 6.8] (I^g = I_E^{-ord_Z I} · I*).

        Equations
        Instances For
          @[reducible, inline]

          Hironaka's weak transform of J by the monoidal transformation with centre D [Hir64, Ch. 0, §5, p. 142; Ch. I, §3, p. 176]: for D non-singular irreducible and m = ν(J_x) at the generic point x of D, f⁻¹(J) is divisible by the m-th power of the ideal of f⁻¹(D), and the weak transform is the quotient.

          Relation to the source.

          • Interpretation. For a disconnected centre D the weak transform divides the pull-back of J by a single power of the whole exceptional divisor: AlgebraicGeometry.Scheme.IdealSheafData.weakTransformAlong is the colon ideal (J.comap π).colon (E ^ c) with c = exceptionalOrderAlong π E J = sSup {c | E ^ c ∣ J.comap π} for the whole exceptional divisor E, that is, the least of the orders along the components. Kollár's componentwise division [Kol07, 58 and Definition 60], stated under $\operatorname{ord}_{D_k} J = d$ for every component $D_k$, agrees with it only under equimultiplicity (the same order along every component of D), which holds wherever this definition enters a main theorem: Main Theorems II and II(N) and Corollary 1 require every centre to be non-singular and irreducible, II(N) and Corollary 1 moreover of constant order on the centre, and the centres of the resolution algorithm lie in the maximal-order locus.
          Equations
          Instances For
            @[reducible, inline]

            The strict (birational) transform of the closed subscheme Y under the blow-up with centre D [Hir64, Ch. 0, §2, pp. 129–130]; [Kol07, Definition 30, 30.2]: AlgebraicGeometry.Scheme.IdealSheafData.strictTransformAlong.

            Equations
            Instances For

              Hironaka's red(f⁻¹(E) ∪ f⁻¹(D)) [Hir64, Main Theorem II (iii); Main Theorem II(N) (3)]: the reduced closed subscheme of the blow-up whose set of points is f⁻¹(E) ∪ f⁻¹(D), as the vanishing ideal sheaf of that closed set (II(N) writes red(f⁻¹(E ∪ D)), the same reduced subscheme: both are the reduced structure on f⁻¹(|E| ∪ |D|)).

              Equations
              Instances For