Finitely many smooth-curve degrees detect numerical triviality
OpenAlgebraicGeometry.RelPicard.exists_finite_curve_degree_tests_of_isProperalgebraic-geometryfinitenessneron-severinumerical-equivalence
Let be a proper scheme over an algebraically closed field . A curve test consists of a smooth proper integral -curve , a -morphism , and two affine open charts covering . For an invertible sheaf on , put
There is a finite set of such tests, depending only on , such that every invertible sheaf satisfies
The degrees are the actual two-chart Cech Euler characteristics. The finite test set is chosen before the line bundle. It may be empty. No smoothness, reducedness, integrality, projectivity, or connectedness of is assumed.
This is a finite-detection consequence of the numerical-rank argument in the cited source, expressed with normalized smooth curves mapping to . It asserts no finite generation modulo algebraic equivalence.
Preamble
import Definitions.Def_AlgebraicGeometry_CurveDegreeTests set_option autoImplicit false open AlgebraicGeometry CategoryTheory universe u
Formal statement
theorem AlgebraicGeometry.RelPicard.exists_finite_curve_degree_tests_of_isProper (K : Type u) [Field K] [IsAlgClosed K] {X : Scheme.{u}}
(π : X ⟶ Spec (.of K)) [IsProper π] :
∃ s : Finset (CurveDegreeTests.Probe π),
∀ M : X.Modules, Scheme.Modules.IsInvertible M →
(∀ t ∈ s, t.degree M = 0) → CurveDegreeTests.NumericallyTrivial π M := by
sorrySource
Cheng, Ji, Larson and Olander, Theorem of the Base (1 July 2021), Lemma 5.4 and Proposition 5.5, pp. 15-16; Lemma 6.1, p. 16; and the numerical-rank reduction in Theorem 7.4, pp. 18-19. https://mattlarson2399.github.io/Papers/theorem-of-the-base.pdf. Finite-dimensionality of the numerical pairing gives a finite detecting subfamily; normalize integral curves and use degree compatibility under proper pullback. This is a derived finite-detection formulation, not a literal source statement. Compare SGA 6, Expose XIII, Proposition 5.2, pp. 650-651.