Masser–Wüstholz — stationary component of a retained equidimensional ideal
ProvedPhilipponMultiplicity.masser_wustholz_stationary_retained_idealAccepted proof-sketch; the terminal retained-cut induction remains Open. The Lean reduction proves stationary-prime selection using charts covering the finitely generated subgroup, dense-open extension of homogeneous vanishing, the retained-prime property, and bounded homogeneous prime avoidance. The sole Open child is refterminal retained hypersurface cut.
Let be a connected commutative algebraic group of dimension over a Philippon base field. Under the original Masser–Wüstholz translation bound and closure-equation bound , set and . Let , , and , and write
Suppose a homogeneous polynomial of degree at most vanishes on but not on all of . For every finitely generated subgroup containing the sampling generators, there exist , homogeneous equations of degree at most , a homogeneous ideal , and a minimal prime of , with the following properties.
Put . Then . Every minimal prime of has Hilbert dimension and its group zero set meets . The ordinary scheme degree satisfies . Every minimal prime of whose group zero set meets is also a minimal prime of . Finally,
The formal statement is unchanged. The remaining child constructs a retained homogeneous ideal of the required dimension and degree, with bounded equations vanishing on the appropriate shrinking grid, at a stage where every permitted next equation belongs to some minimal prime. Stationary-prime selection is proved by contradiction from this terminality. Source: Masser–Wüstholz, Chapter 1 §2, printed p.414, equation (1.3): https://gdz.sub.uni-goettingen.de/id/PPN356556735_0072 .
import Definitions.Def_PhilipponMultiplicity_GeometricSupport import Definitions.Def_PhilipponMultiplicity_SectionFive set_option autoImplicit false open scoped BigOperators
namespace PhilipponMultiplicity
theorem masser_wustholz_stationary_retained_ideal
(K : Type*) [NontriviallyNormedField K] (hK : IsPhilipponBaseField K)
(E : EmbeddedCommutativeGroup K)
(hn : 0 < (singleGroupProduct E).dimension)
(hconnected : @_root_.IsConnected _ (singleGroupProduct E).zariskiTopology Set.univ)
(a b : ℕ) (ha : 1 ≤ a) (hb : 1 ≤ b)
(htranslation : MWTranslationBound (singleGroupProduct E) a)
(hclosure : ∃ equations : Finset (singleGroupProduct E).CoordinateRing,
(∀ P ∈ equations, (singleGroupProduct E).ambient.IsHomogeneousAtMost P (fun _ => b)) ∧
groupProjectiveClosure (singleGroupProduct E) =
{x | ∀ P ∈ equations, (singleGroupProduct E).ambient.eval P x = 0})
(m D : ℕ) (hm : 1 ≤ m) (hD : 1 ≤ D)
(γ : Fin m → (singleGroupProduct E).Point) (R : ℝ) (hR : 1 ≤ R)
(P : (singleGroupProduct E).CoordinateRing)
(hP : (singleGroupProduct E).ambient.IsHomogeneousAtMost P (fun _ => D)) :
let G := singleGroupProduct E
let c : ℝ := 1 / ((a : ℝ) ^ G.dimension * (b : ℝ) ^ (E.ambientDimension - G.dimension))
(∀ x ∈ samplingGrid γ ((G.dimension : ℝ) * R),
G.ambient.eval P (G.embedding x) = 0) →
(∃ x : G.Point, G.ambient.eval P (G.embedding x) ≠ 0) →
∀ Γ : Submodule ℤ G.Point, Γ.FG → (∀ i, γ i ∈ Γ) →
∃ r : ℕ, 1 ≤ r ∧ r ≤ G.dimension ∧
∃ F : Finset G.CoordinateRing,
(∀ Q ∈ F, ∃ d : G.FactorIndex → ℕ,
G.ambient.IsHomogeneous Q d ∧
∀ i, (d i : ℝ) ≤ (a : ℝ) ^ (G.dimension - 1) * (D : ℝ)) ∧
∃ J : Ideal G.CoordinateRing, IsMultihomogeneousIdeal G.ambient J ∧
(G.vanishingIdeal Set.univ ⊔ Ideal.span (F : Set G.CoordinateRing) ≤ J) ∧
(∀ q ∈ J.minimalPrimes,
SectionThree.idealDimension G.ambient q = SectionThree.idealDimension G.ambient J) ∧
SectionThree.idealDimension G.ambient J = G.dimension - r ∧
((SectionThree.idealDegreeValue G.ambient J (fun _ => 1) : ℚ) : ℝ) ≤
((D : ℝ) / c) ^ r ∧
(∀ q ∈ J.minimalPrimes, ∃ x ∈ Γ, x ∈ idealZeroLocusOnGroup G q) ∧
(∀ q ∈ (G.vanishingIdeal Set.univ ⊔ Ideal.span (F : Set G.CoordinateRing)).minimalPrimes,
(∃ x ∈ Γ, x ∈ idealZeroLocusOnGroup G q) → q ∈ J.minimalPrimes) ∧
∃ p ∈ J.minimalPrimes, ∀ g ∈ samplingGrid γ R,
translate g (idealZeroLocusOnGroup G p) ⊆ idealZeroLocusOnGroup G J := by sorry
end PhilipponMultiplicity