Coordinate projections: analytic contact and subgroup obstructions
ProvedPhilipponMultiplicity.coordinate_projection_obstruction_transportAccepted proof-sketch; only tangent-kernel transport remains Open. The Lean reduction explicitly constructs the projected analytic parametrization on the original parameter ball and proves carrier equality and contact-order preservation. It constructs actual algebraic subgroup preimages, proves inheritance of disjoint factors, identifies selected factor codimensions, and proves equality of sampling quotient ranks. Its sole Open child is reftangent kernels of subgroup preimages.
Let be an embedded product of commutative algebraic groups over a Philippon base field, with disjoint factors. Choose a nonempty set of its factors, keep their original embeddings, and let
be the coordinate projection. For an exponent tuple on , write for its extension by zero outside . Then has disjoint factors, and every analytic subgroup of admits an analytic subgroup of with
For every multihomogeneous polynomial on and every , its actual pullback by coordinate renaming satisfies
For each algebraic subgroup , there is an algebraic subgroup whose carrier is exactly . With on , it satisfies
For every finite list of sampling generators , the corresponding quotient ranks are equal:
The codimensions use the existing Hilbert dimensions of factor projections and the existing tangent-kernel definition for analytic subgroups. The rank is the rank of the subgroup generated in the actual quotient. No injectivity or minimality of the analytic parametrization is required.
This is the remaining geometric input in a reduction of the zero-degree branch of Philippon's Corollary 2.3. It transports individual obstruction subgroups, so it makes no assertion comparing empty-family infima. The polynomial support reduction, grid lifting, uniform choice of the numerical constant and multiplicity-estimate argument are separate proved parts of the parent sketch. This statement assumes no multiplicity estimate or vanishing conclusion.
The formal statement is unchanged. The remaining inclusion is between tangent kernels defined from the actual vanishing ideals on the same parameter space. The analytic-codimension inequality is proved from that inclusion by finite-dimensional rank monotonicity. No additional hypothesis is imposed on the original theorem.
import Definitions.Def_PhilipponMultiplicity_FactorProjection set_option autoImplicit false open scoped BigOperators Topology
namespace PhilipponMultiplicity
/-- Coordinate projection transports analytic contact and the actual
algebraic subgroup obstructions. No multiplicity estimate is assumed. -/
theorem coordinate_projection_obstruction_transport
(K : Type*) [NontriviallyNormedField K] (hK : IsPhilipponBaseField K)
(G : EmbeddedGroupProduct K) (hdisjoint : HasDisjointFactors G)
(s : GroupFactorSelection G) :
HasDisjointFactors s.group ∧
∀ A : AnalyticSubgroup G, ∃ B : AnalyticSubgroup s.group,
B.carrier = s.project '' A.carrier ∧
(∀ (Q : s.group.CoordinateRing) (D : s.group.FactorIndex → ℕ),
IsMultihomogeneousOfDegree s.group Q D → ∀ g : G.Point,
vanishingOrder A (MvPolynomial.rename s.coordinateIndex Q) g =
vanishingOrder B Q (s.project g)) ∧
(∀ H : AlgebraicSubgroup s.group, ∃ H' : AlgebraicSubgroup G,
H'.carrier = s.project ⁻¹' H.carrier ∧
(∀ i, s.extendExponent (factorCodimension s.group H) i ≤
factorCodimension G H' i) ∧
analyticCodimension A H'.carrier ≤ analyticCodimension B H.carrier ∧
∀ (l : ℕ) (γ : Fin l → G.Point),
samplingQuotientRank γ H' = samplingQuotientRank (fun i => s.project (γ i)) H) := by sorry
end PhilipponMultiplicity