Intersection numbers from Snapper coefficients #
Intersection numbers are the top mixed finite differences of the Euler characteristic of line-bundle twists. No cycle group or Chow ring occurs in the construction.
The geometric input remains explicit. TwistContext records coherent representatives and the
Snapper.GeometricInduction certificates for every finite family of Picard classes, together
with one Picard-level Euler function which they realize. IntersectionContext specializes the
base coherent sheaf to the structure sheaf. Thus the missing geometric construction is data in
a theorem statement, never a global axiom.
Finite differences on the Picard group #
One finite difference of an integer-valued function on the Picard group.
Equations
- AlgebraicGeometry.IntersectionTheory.Number.picardDifference L f M = f (M * L) - f M
Instances For
Successive finite differences in a list of Picard classes.
Equations
- One or more equations did not get rendered due to their size.
- AlgebraicGeometry.IntersectionTheory.Number.picardMixedDifference [] x✝ = x✝
Instances For
A mixed Picard coefficient is evaluated at the trivial line bundle.
Equations
Instances For
Picard mixed differences are symmetric in their directions.
Splitting a tensor-product direction introduces one second-difference correction.
Picard-group degree at most d, in the coefficient form needed for intersections.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Additivity of a top Picard coefficient in its first class.
Additivity of a top Picard coefficient in any class slot.
With every other class fixed, a top Picard coefficient is an additive homomorphism from the Picard group written additively.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Integer homogeneity of a top Picard coefficient in every slot.
The Picard class represented by an integer vector in a finite family.
Equations
- AlgebraicGeometry.IntersectionTheory.Number.picardMonomial L n = ∏ i : ι, L i ^ n i
Instances For
Restrict a Picard-level function to the exponent lattice generated by L.
Equations
Instances For
Lattice mixed differences are Picard mixed differences in the represented classes.
The standard-coordinate top coefficient is the intrinsic Picard mixed coefficient.
Explicit geometric contexts #
Uniform Snapper data for twists of one coherent sheaf by every finite family of line
bundles. The single eulerPic field makes coefficients independent of the chosen finite
presentation; realization identifies it with the actual geometric Euler characteristic.
- eulerPic : Scheme.Modules.Pic X → ℤ
- realization (r : ℕ) (L : Fin r → Scheme.Modules.Pic X) (n : NumericalPolynomial.Lattice (Fin r)) : self.eulerPic (picardMonomial L n) = Snapper.eulerFunction D ⋯ n
Instances For
The structure sheaf as a coherent object, from the explicitly supplied coherence proof.
Equations
- AlgebraicGeometry.IntersectionTheory.Number.structureSheafObject h = { obj := SheafOfModules.unit X.ringCatSheaf, property := h }
Instances For
Snapper input specialized to twists of the structure sheaf.
- structureSheafCoherent : Scheme.Modules.IsCoherent X (SheafOfModules.unit X.ringCatSheaf)
- twists : TwistContext D (structureSheafObject ⋯) d
Instances For
The geometric Euler function of any chosen finite presentation is the restriction of the intrinsic Picard-level Euler function.
Snapper polynomiality for every finite presentation implies intrinsic Picard coefficient
vanishing in degree d + 1.
The intrinsic Euler characteristic of a Picard class.
Instances For
The Picard Euler function has finite-difference degree at most the geometric dimension.
The dimension-d intersection number of d Picard classes, defined as the top mixed
Snapper coefficient of the corresponding structure-sheaf twists.
Equations
Instances For
Intrinsic formulation: the intersection number is the iterated Picard difference of
eulerPic, evaluated at the trivial bundle.
List form of the intersection coefficient. The separate list API is convenient for multilinearity proofs and for downstream Chern-character formulas.
Equations
- P.picardIntersectionList classes = AlgebraicGeometry.IntersectionTheory.Number.picardCoefficient classes P.eulerPic
Instances For
Intersection numbers are invariant under line-bundle isomorphism. Equality in Pic X is
exactly equality of isomorphism classes.
Symmetry in all divisor arguments.
Multilinearity, in list form: tensor product in any one slot becomes addition of intersection numbers.
Integer homogeneity, in list form.
A trivial line-bundle argument makes every positive-arity intersection vanish.
Cartier divisors and divisor classes #
Intersection numbers of Cartier divisor classes, through the additive class-to-Picard map.
Equations
- P.cartierClassIntersectionNumber classes = P.picardIntersectionNumber fun (i : Fin d) => Additive.toMul (AlgebraicGeometry.Scheme.CartierDivisor.classToPicAdd (classes i))
Instances For
Intersection numbers of actual Cartier divisors depend only on their divisor classes.
Equations
- P.cartierDivisorIntersectionNumber divisors = P.cartierClassIntersectionNumber fun (i : Fin d) => (AlgebraicGeometry.Scheme.CartierDivisor.toClass X) (divisors i)
Instances For
The divisor and Picard formulations agree under D ↦ O_X(D).
Principal equivalence does not change intersection numbers.
Adding a principal divisor in every slot leaves the intersection number unchanged.
Symmetry also holds in the Cartier-divisor formulation.
Dimension-zero, curve, and surface specializations #
In dimension zero the empty intersection is honestly the empty coefficient: χ(O_X).
On a curve the divisor number is the first difference χ(L) - χ(O_X).
The binary surface intersection coefficient.
Equations
Instances For
The binary name is exactly the Fin 2 top Snapper coefficient.
The surface coefficient is the expected inclusion-exclusion expression.
The surface specialization as a symmetric bilinear form on the additive Picard group.
Equations
- One or more equations did not get rendered due to their size.