Prove2Me
Navigate
DiscoverFormalpediaBlogsUsersMomentumMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

Proposition 3.3 — multigraded intersection bounds

Proved
PhilipponMultiplicity.proposition_3_3

by tomasz · Sep 23, 2026 · Mathlib 0df444a (Lean v4.33.1)

draft-statementphilippon-multiplicity

Proved with no Open theorem dependencies, verified 29 September 2026. Checked locally with Lean 4.33.1 and the proposal’s pinned Mathlib. An independent blind readback is attached.

Adding multihomogeneous equations of the prescribed degree bounds does not increase the source sum of radical component Hilbert forms. On an open maximal-spectrum locus where the original quotient is locally Cohen–Macaulay, the corresponding inequality retains primary component multiplicities. Both inequalities are required.

Preamble
/-
Open statement draft. The proof and source-comparison obligations remain open.
-/
import Definitions.Def_PhilipponMultiplicity_Degree

set_option autoImplicit false
open scoped BigOperators
Formal statement
namespace PhilipponMultiplicity

theorem proposition_3_3
    (K : Type*) [NontriviallyNormedField K] (hK : IsPhilipponBaseField K)
    (M : MultiProjectiveSpace K)
    (I₀ : Ideal M.CoordinateRing) (hI₀ : IsMultihomogeneousIdeal M I₀)
    (m : ℕ) (P : Fin m → M.CoordinateRing) (D : M.FactorIndex → ℕ)
    (hP : ∀ j, IsMultihomogeneousOfDegreeAtMost M (P j) D) :
    let I := I₀ ⊔ Ideal.span (Set.range P)
    (componentHilbertSum M I.radical (⊤ : MaximalOpenLocus M) D ≤
      componentHilbertSum M I₀.radical (⊤ : MaximalOpenLocus M) D) ∧
    (∀ U : MaximalOpenLocus M, IsLocallyCohenMacaulayOn M I₀ U →
      componentHilbertSum M I U D ≤ componentHilbertSum M I₀ U D) := by sorry

end PhilipponMultiplicity
Source
1986, pp. 365–370. https://numdam.org/articles/10.24033/bsmf.2060/
Read-back

What the Lean code literally says, in plain math · gpt-6

Let KKK be any nontrivially normed field for which there is an isometric field isomorphism either K≅CK\cong\mathbb CK≅C, or K≅CpK\cong\mathbb C_pK≅Cp​ for some prime natural number ppp, where Cp\mathbb C_pCp​ is the field denoted by PadicComplex(p)\mathrm{PadicComplex}(p)PadicComplex(p). For every product MMM of a positive number sss of projective spaces of dimensions Ni∈NN_i\in\mathbb NNi​∈N, every multihomogeneous ideal I0I_0I0​ in its coordinate ring RRR, every m∈Nm\in\mathbb Nm∈N, every list (Pj)j=1m(P_j)_{j=1}^m(Pj​)j=1m​ of polynomials, and every natural block-degree vector DDD, suppose each PjP_jPj​ is homogeneous of some block degree eje_jej​ with (ej)i≤Di(e_j)_i\leq D_i(ej​)i​≤Di​ for every iii. Let I=I0+(P1,…,Pm)I=I_0+(P_1,\ldots,P_m)I=I0​+(P1​,…,Pm​). Then both of the following statements hold: SMaxSpec⁡R(I,D)≤SMaxSpec⁡R(I0,D)S_{\operatorname{MaxSpec}R}(\sqrt I,D)\leq S_{\operatorname{MaxSpec}R}(\sqrt{I_0},D)SMaxSpecR​(I​,D)≤SMaxSpecR​(I0​​,D); and, for every open subset UUU of the maximal-ideal spectrum of RRR, if I0I_0I0​ satisfies the local condition below at every point of UUU, then SU(I,D)≤SU(I0,D)S_U(I,D)\leq S_U(I_0,D)SU​(I,D)≤SU​(I0​,D). These are inequalities of rational numbers. In the first inequality the component ideals are formed from the radicals; in the second they are formed from the original ideals. For this space let R=K[Xij:1≤i≤s, 0≤j≤Ni]R=K[X_{ij}:1\leq i\leq s,\ 0\leq j\leq N_i]R=K[Xij​:1≤i≤s, 0≤j≤Ni​]. An ideal is multihomogeneous here precisely when it contains every block-homogeneous component of every one of its elements. Write Birr=⋂i(Xi0,…,XiNi)B_{\mathrm{irr}}=\bigcap_i(X_{i0},\ldots,X_{iN_i})Birr​=⋂i​(Xi0​,…,XiNi​​). For an ideal JJJ, a minimal prime q\mathfrak qq is counted on an open set UUU of the maximal-ideal spectrum exactly when Birr⊈qB_{\mathrm{irr}}\nsubseteq\mathfrak qBirr​⊈q and some maximal ideal m∈U\mathfrak m\in Um∈U contains q\mathfrak qq. Its contributing ideal is CJ,q=(JRq)∩RC_{J,\mathfrak q}=(JR_{\mathfrak q})\cap RCJ,q​=(JRq​)∩R, obtained by extending to the localization at q\mathfrak qq and contracting. For any ideal LLL, form the rational polynomial QLQ_LQL​ which equals the dimension of the image of the degree-eee homogeneous piece in R/LR/LR/L for every coordinatewise sufficiently large e∈Nse\in\mathbb N^se∈Ns, with QL=0Q_L=0QL​=0 if such a polynomial does not exist. Put dL=deg⁡QLd_L=\deg Q_LdL​=degQL​, taking deg⁡0=0\deg 0=0deg0=0, and FL(D)=dL! [QL]dL(D)F_L(D)=d_L!\,[Q_L]_{d_L}(D)FL​(D)=dL​![QL​]dL​​(D). The rational number SU(J,D)S_U(J,D)SU​(J,D) is the finite sum of FCJ,q(D)F_{C_{J,\mathfrak q}}(D)FCJ,q​​(D) over precisely the counted minimal primes. It does not use an externally supplied component list or multiplicity; components of all dimensions are included when they pass the two tests. For an ideal I0⊆RI_0\subseteq RI0​⊆R and a maximal ideal m\mathfrak mm, the local condition on I0I_0I0​ means that B=Rm/I0RmB=R_{\mathfrak m}/I_0R_{\mathfrak m}B=Rm​/I0​Rm​ is the zero ring, or there exists a finite list of nonunits in BBB forming a regular sequence on BBB whose length, as an element of N∪{∞,−∞}\mathbb N\cup\{\infty,-\infty\}N∪{∞,−∞}, equals the Krull dimension of BBB. Regularity means that multiplication by each entry is injective after quotienting by the previous entries and that the final quotient is nonzero. The condition on UUU requires this at every maximal ideal in UUU. No nonzero assumption is imposed on the PjP_jPj​, no common exact degree is required, and no properness assumption is imposed on I0I_0I0​ or III. The case m=0m=0m=0 gives I=I0I=I_0I=I0​ and both inequalities are equalities. For U=∅U=\varnothingU=∅ the local hypothesis is vacuous and both component sums are zero. An ideal with no minimal primes contributes an empty sum, hence zero; in particular this holds for the unit ideal. Zero entries of DDD are allowed, with 00=10^0=100=1, and the zero fallback for a Hilbert polynomial contributes zero. A zero-dimensional projective factor is allowed; a product with no factors is not.

Human review
  • Endorsed by Shuze Chen · Sep 29, 2026

    Confirmed by the moderator at approval.

  • Endorsed by tomasz · Sep 29, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paper
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me