Primary multiplicities at the endpoint of dimension descent
ProvedPhilipponMultiplicity.SectionThreeSupport.componentSum_le_of_discarded_zeroalgebraic-geometrycommutative-algebra
Let be multihomogeneous ideals in the coordinate ring of a product of projective spaces over any field. Assume that every relevant minimal prime of which is not an associated prime of has Hilbert dimension zero. For every open set of maximal ideals and every natural degree vector , the sum of the degree forms of the actual isolated primary components of meeting is at most . These are the canonical localized primary components, so the comparison retains their scheme multiplicities. Zero entries in are allowed. This proves the final comparison in the descent, rather than the estimates at its intermediate cuts.
Preamble
import Definitions.Def_PhilipponMultiplicity_SectionThreeSupport set_option autoImplicit false open scoped BigOperators open MvPolynomial PhilipponMultiplicity PhilipponMultiplicity.SectionThree open PhilipponMultiplicity.SectionThreeSupport PhilipponMultiplicity.Hilbert
Formal statement
theorem PhilipponMultiplicity.SectionThreeSupport.componentSum_le_of_discarded_zero
{K : Type*} [Field K] (M : MultiProjectiveSpace K)
(J I : Ideal M.CoordinateRing)
(hJ : IsMultihomogeneousIdeal M J) (hI : IsMultihomogeneousIdeal M I)
(hJI : J ≤ I)
(hbound : ∀ q ∈ J.minimalPrimes,
Hilbert.IsRelevant K M.factorCount M.ambientDimension q →
q ∉ associatedPrimes M.CoordinateRing (M.CoordinateRing ⧸ I) →
idealDimension M q ≤ 0)
(U : MaximalOpenLocus M) (D : M.FactorIndex → ℕ) :
componentHilbertSum M I U D ≤ componentHilbertSum M J U D := by sorrySource
Philippon (1986), Lemmes de zéros dans les groupes algébriques commutatifs. The scheme-theoretic endpoint comparison in the proof of Proposition 3.3, printed p. 366. The discarded-prime zero-dimensional bound is the concrete final-stage invariant. The inequalities along the cuts remain separate obligations. https://www.numdam.org/articles/10.24033/bsmf.2060/