Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Invariant derivations, translations, zero sets and stabilisers on GaA0×GmA1\mathbb{G}_a^{A_0} \times \mathbb{G}_m^{A_1}GaA0​​×GmA1​​, for the multiplicity estimate

Definition
DiazMultiplicity

by carlok · Oct 30, 2026 · Mathlib 0df444a (Lean v4.33.1)

commutative-algebradiaz-modulus-leanmultiplicity-estimatetranscendence

Vocabulary for the multiplicity side of the elementary multiplicity estimate MultiplicityEstimateElem, used by the C1.3 nodes of the Diaz modulus mission. It builds on the published Definition DiazHilbertBox (module Definitions.Def_DiazHilbertBox): everything lives in the polynomial ring R=K[X,Y,Z]R = K[X, Y, Z]R=K[X,Y,Z] = MvPolynomial (A₀ ⊕ (A₁ × Bool)) K with variables XhX_hXh​ (h∈A0h \in A_0h∈A0​), YiY_iYi​ = inr (i, true) and ZiZ_iZi​ = inr (i, false) (i∈A1i \in A_1i∈A1​), and J0=(YiZi−1)iJ_0 = (Y_iZ_i - 1)_iJ0​=(Yi​Zi​−1)i​ is torusIdeal. It contains fifteen definitions and five short structural lemmas, and no theorem of substance.

The group. TorusGroup K A₀ A₁ is G=GaA0×GmA1G = \mathbb{G}_a^{A_0} \times \mathbb{G}_m^{A_1}G=GaA0​​×GmA1​​ over KKK, the pairs g=(x,y)g = (x, y)g=(x,y) with x∈KA0x \in K^{A_0}x∈KA0​ (written multiplicatively) and y∈(K×)A1y \in (K^\times)^{A_1}y∈(K×)A1​; torusPt g is the point (x,y,y−1)(x, y, y^{-1})(x,y,y−1).

Invariant derivations. For w=(ξ,η)∈KA0×KA1w = (\xi, \eta) \in K^{A_0} \times K^{A_1}w=(ξ,η)∈KA0​×KA1​, torusDeriv w is the derivation

Dw=∑hξh∂∂Xh+∑iηi(Yi∂∂Yi−Zi∂∂Zi)D_w = \sum_h \xi_h \frac{\partial}{\partial X_h} + \sum_i \eta_i \Bigl(Y_i \frac{\partial}{\partial Y_i} - Z_i \frac{\partial}{\partial Z_i}\Bigr)Dw​=h∑​ξh​∂Xh​∂​+i∑​ηi​(Yi​∂Yi​∂​−Zi​∂Zi​∂​)

of RRR, the lift of DALAG's invariant derivation (Lemma 8.11); torusDerivList l is Dw1∘⋯∘DwtD_{w_1} \circ \cdots \circ D_{w_t}Dw1​​∘⋯∘Dwt​​ for a list l=[w1,…,wt]l = [w_1, \dots, w_t]l=[w1​,…,wt​].

Translations, ideals of points, zero sets. torusTransl g is the KKK-algebra map Xh↦Xh+xhX_h \mapsto X_h + x_hXh​↦Xh​+xh​, Yi↦yiYiY_i \mapsto y_i Y_iYi​↦yi​Yi​, Zi↦yi−1ZiZ_i \mapsto y_i^{-1} Z_iZi​↦yi−1​Zi​, i.e. f↦f∘τgf \mapsto f \circ \tau_gf↦f∘τg​; torusVanishing S is the ideal I(S)I(S)I(S) of polynomials vanishing at every point of S⊆GS \subseteq GS⊆G; torusZero I is the zero set ZG(I)⊆GZ_G(I) \subseteq GZG​(I)⊆G of an ideal; torusStab I is the stabiliser {g:ZG(I) g⊆ZG(I)}\{g : Z_G(I)\,g \subseteq Z_G(I)\}{g:ZG​(I)g⊆ZG​(I)}, as a set.

Subgroups and sections. torusMono φ is the Laurent monomial Yφ+Zφ−Y^{\varphi^+} Z^{\varphi^-}Yφ+Zφ− for φ∈ZA1\varphi \in \mathbb{Z}^{A_1}φ∈ZA1​ (the character yφy^\varphiyφ on GGG); linForm a is ∑hahXh\sum_h a_h X_h∑h​ah​Xh​; standardSet V φ is {(x,y):x∈V, yφk=1 for every k}\{(x, y) : x \in V,\ y^{\varphi_k} = 1 \text{ for every } k\}{(x,y):x∈V, yφk​=1 for every k}, the subgroup V×TφV \times T_\varphiV×Tφ​; and sectionIdeal ℓ ψ is J0+(∑hℓk,hXh)k+(Yψk+Zψk−−1)kJ_0 + (\sum_h \ell_{k,h} X_h)_k + (Y^{\psi_k^+} Z^{\psi_k^-} - 1)_kJ0​+(∑h​ℓk,h​Xh​)k​+(Yψk+​Zψk−​−1)k​.

Vanishing and the descent. vanishOrd J W t is the set of Q∈RQ \in RQ∈R with DlQ∈JD_l Q \in JDl​Q∈J for every list lll of elements of the subspace WWW of length at most ttt (for J=I(S)J = I(S)J=I(S): QQQ vanishes to order >t> t>t along WWW on SSS). hmePoint u a s is the point gs=(∑jsjuj,(∏jajisj)i)g_s = (\sum_j s_j u_j, (\prod_j a_{ji}^{s_j})_i)gs​=(∑j​sj​uj​,(∏j​ajisj​​)i​) of the multiplicity estimate. obstructionSpan P W u a m n V is the span of the (Dw1⋯DwtP)∘τgsv(D_{w_1} \cdots D_{w_t} P) \circ \tau_{g_s v}(Dw1​​⋯Dwt​​P)∘τgs​v​ with ∣sj∣≤m|s_j| \le m∣sj​∣≤m, v∈Vv \in Vv∈V, t≤nt \le nt≤n and wi∈Ww_i \in Wwi​∈W: with V={1}V = \{1\}V={1} it spans over J0J_0J0​ the ideals IrI_rIr​ of the descent of DALAG §8.4.1, and with V=ZG(p)V = Z_G(\mathfrak{p})V=ZG​(p) the ideal AAA there.

Structural lemmas. The values of DwD_wDw​ on the variables (DwXh=ξhD_w X_h = \xi_hDw​Xh​=ξh​, DwYi=ηiYiD_w Y_i = \eta_i Y_iDw​Yi​=ηi​Yi​, DwZi=−ηiZiD_w Z_i = -\eta_i Z_iDw​Zi​=−ηi​Zi​), torusDerivList [] = 1, the membership rule of torusZero, and eval_torusTransl: the value of σgf\sigma_g fσg​f at the point of hhh is the value of fff at the point of hghghg.

Definition code
import Mathlib
import Definitions.Def_DiazHilbertBox

/-! # The group `𝔾ₐ^{A₀} × 𝔾ₘ^{A₁}`, its invariant derivations, translations, zero sets and stabilisers

The vocabulary of the multiplicity side of the elementary multiplicity estimate (DALAG Ch. 8, §8.3 and §8.4.1,
at finite level). Everything lives in the model `R = K[X, Y, Z] = MvPolynomial (A₀ ⊕ (A₁ × Bool)) K` of
`Def_DiazHilbertBox` (`Y_i = inr (i, true)`, `Z_i = inr (i, false)`), with `J₀ = torusIdeal K A₀ A₁`.

* `TorusGroup K A₀ A₁`: `G = 𝔾ₐ^{A₀} × 𝔾ₘ^{A₁}`, written multiplicatively; `torusPt g = (x, y, y⁻¹)`.
* `torusDeriv w`: the invariant derivation `D_w` (DALAG Lemma 8.11), `w = (ξ, η)`, lifted to `R`;
  `torusDerivList l = D_{w₁} ∘ ⋯ ∘ D_{w_t}` for a list `l = [w₁, …, w_t]`.
* `torusTransl g`: the translation `f ↦ f ∘ τ_g`, with `eval_torusTransl`.
* `torusVanishing S`, `torusZero I`: the ideal `I(S)` of a set of points and the zero set `Z_G(I)` of an ideal.
* `torusStab I = {g : Z(I) · g ⊆ Z(I)}`: the stabiliser of the zero set, as a set.
* `torusMono φ = Y^{φ⁺} Z^{φ⁻}`, `linForm a = Σ a_h X_h`, `standardSet V φ = V × T_φ`.
* `sectionIdeal ℓ ψ = J₀ + (Σ_h ℓ_k h X_h)_k + (Y^{ψ_k⁺}Z^{ψ_k⁻} − 1)_k`.
* `vanishOrd J W t`: vanishing to order `> t` along `W`, relative to the ideal `J`.
* `hmePoint u a s = g_s = (Σ_j s_j u_j, Π_j a_j^{s_j})`, and `obstructionSpan`: with `V = {1}` it spans the
  descent ideal `I_r` over `J₀`, with `V = Z(𝔭)` the ideal `A` of DALAG §8.4.1.

Only definitions and structural lemmas (each a few lines) live here. -/

namespace DiazModulus

open MvPolynomial

/-- The group `G = 𝔾ₐ^{A₀} × 𝔾ₘ^{A₁}` over `K`, written multiplicatively. -/
abbrev TorusGroup (K A₀ A₁ : Type*) [Field K] := Multiplicative (A₀ → K) × (A₁ → Kˣ)

/-- The point `(x, y, y⁻¹)` of `K^{A₀ ⊕ (A₁ × Bool)}` over `g = (x, y) ∈ G`. -/
def torusPt {K A₀ A₁ : Type*} [Field K] (g : TorusGroup K A₀ A₁) : A₀ ⊕ (A₁ × Bool) → K :=
  torusPoint (Multiplicative.toAdd g.1) g.2

/-- DALAG's invariant derivation `D_w` (Lemma 8.11), lifted to `K[X, Y, Z]`:
`Σ_h ξ_h ∂/∂X_h + Σ_i η_i (Y_i ∂/∂Y_i − Z_i ∂/∂Z_i)` for `w = (ξ, η)`. -/
noncomputable def torusDeriv {K A₀ A₁ : Type*} [Field K] (w : (A₀ → K) × (A₁ → K)) :
    Derivation K (MvPolynomial (A₀ ⊕ (A₁ × Bool)) K) (MvPolynomial (A₀ ⊕ (A₁ × Bool)) K) :=
  mkDerivation K (Sum.elim (fun h => C (w.1 h)) fun p : A₁ × Bool =>
    if p.2 then C (w.2 p.1) * X (Sum.inr p) else -(C (w.2 p.1) * X (Sum.inr p)))

/-- `D_{w₁} ∘ ⋯ ∘ D_{w_t}` for the list `[w₁, …, w_t]`. -/
noncomputable def torusDerivList {K A₀ A₁ : Type*} [Field K]
    (l : List ((A₀ → K) × (A₁ → K))) : Module.End K (MvPolynomial (A₀ ⊕ (A₁ × Bool)) K) :=
  (l.map fun w => (torusDeriv w).toLinearMap).prod

/-- Translation `f ↦ f ∘ τ_g` (`X ↦ X + x`, `Y ↦ y Y`, `Z ↦ y⁻¹ Z`); an `AlgEquiv` by `DiazModulus.exists_translation_algEquiv`. -/
noncomputable def torusTransl {K A₀ A₁ : Type*} [Field K] (g : TorusGroup K A₀ A₁) :
    MvPolynomial (A₀ ⊕ (A₁ × Bool)) K →ₐ[K] MvPolynomial (A₀ ⊕ (A₁ × Bool)) K :=
  aeval (Sum.elim (fun h => X (Sum.inl h) + C (Multiplicative.toAdd g.1 h))
    (fun p => C (torusPt g (Sum.inr p)) * X (Sum.inr p)))

/-- The ideal `I(S)` of a set of points of `G`, in `K[X, Y, Z]`. -/
noncomputable def torusVanishing {K A₀ A₁ : Type*} [Field K] (S : Set (TorusGroup K A₀ A₁)) :
    Ideal (MvPolynomial (A₀ ⊕ (A₁ × Bool)) K) :=
  ⨅ g ∈ S, RingHom.ker (eval (torusPt g))

/-- The zero set `Z_G(I)` in `G` of an ideal of `K[X, Y, Z]`. -/
def torusZero {K A₀ A₁ : Type*} [Field K] (I : Ideal (MvPolynomial (A₀ ⊕ (A₁ × Bool)) K)) :
    Set (TorusGroup K A₀ A₁) :=
  {g | ∀ f ∈ I, eval (torusPt g) f = 0}

/-- The stabiliser `{g ∈ G : Z(I) · g ⊆ Z(I)}` of the zero set of `I` (DALAG §8.4.1, "as in §5.4"). -/
def torusStab {K A₀ A₁ : Type*} [Field K] (I : Ideal (MvPolynomial (A₀ ⊕ (A₁ × Bool)) K)) :
    Set (TorusGroup K A₀ A₁) :=
  {g | ∀ z ∈ torusZero I, z * g ∈ torusZero I}

/-- The Laurent monomial `Y^{φ⁺} Z^{φ⁻}` (the character `y^φ` on `G`). -/
noncomputable def torusMono {K A₀ A₁ : Type*} [Field K] [Fintype A₁] (φ : A₁ → ℤ) :
    MvPolynomial (A₀ ⊕ (A₁ × Bool)) K :=
  ∏ i, X (Sum.inr (i, true)) ^ (φ i).toNat * X (Sum.inr (i, false)) ^ (-φ i).toNat

/-- The linear form `Σ_h a_h X_h`. -/
noncomputable def linForm {K A₀ A₁ : Type*} [Field K] [Fintype A₀] (a : A₀ → K) :
    MvPolynomial (A₀ ⊕ (A₁ × Bool)) K :=
  ∑ h, C (a h) * X (Sum.inl h)

/-- The algebraic subgroup `V × T_φ` of `G`, with `T_φ = {y : y^{φ k} = 1 for all k}`. -/
def standardSet {K A₀ A₁ : Type*} [Field K] [Fintype A₁] (V : Submodule K (A₀ → K)) {ι : Type*}
    (φ : ι → A₁ → ℤ) : Set (TorusGroup K A₀ A₁) :=
  {g | Multiplicative.toAdd g.1 ∈ V ∧ ∀ k, ∏ i, g.2 i ^ φ k i = 1}

/-- The section ideal `𝔞 = J₀ + (Σ_h ℓ_k h X_h)_k + (Y^{ψ_k⁺} Z^{ψ_k⁻} − 1)_k`. -/
noncomputable def sectionIdeal {K A₀ A₁ ι κ : Type*} [Field K] [Fintype A₀] [Fintype A₁]
    (ℓ : ι → A₀ → K) (ψ : κ → A₁ → ℤ) : Ideal (MvPolynomial (A₀ ⊕ (A₁ × Bool)) K) :=
  torusIdeal K A₀ A₁ ⊔ Ideal.span (Set.range fun k => linForm (A₁ := A₁) (ℓ k))
    ⊔ Ideal.span (Set.range fun k => torusMono (A₀ := A₀) (ψ k) - 1)

/-- Vanishing to order `> t` along `W`, relative to the ideal `J`: every `D_{w₁}⋯D_{w_s} Q` with
`s ≤ t`, `wᵢ ∈ W`, lies in `J` (for `J = I(S)`: `Q` vanishes to order `> t` along `W` on `S`). -/
def vanishOrd {K A₀ A₁ : Type*} [Field K]
    (J : Ideal (MvPolynomial (A₀ ⊕ (A₁ × Bool)) K)) (W : Submodule K ((A₀ → K) × (A₁ → K)))
    (t : ℕ) : Set (MvPolynomial (A₀ ⊕ (A₁ × Bool)) K) :=
  {Q | ∀ l : List ((A₀ → K) × (A₁ → K)), l.length ≤ t → (∀ w ∈ l, w ∈ W) →
    torusDerivList l Q ∈ J}

/-- The point `g_s = (Σ_j s_j u_j, Π_j a_j^{s_j})` of the multiplicity estimate. -/
noncomputable def hmePoint {K A₀ A₁ B₁ : Type*} [Field K] [Fintype B₁] (u : B₁ → A₀ → K)
    (a : B₁ → A₁ → Kˣ) (s : B₁ → ℤ) : TorusGroup K A₀ A₁ :=
  (Multiplicative.ofAdd (∑ j, (s j : K) • u j), fun i => ∏ j, a j i ^ s j)

/-- The span of §8.4.1's family: `(D_{w₁}⋯D_{w_t} P) ∘ τ_{g_s · v}` for `|s_j| ≤ m`, `v ∈ V`,
`t ≤ n`, `wᵢ ∈ W`. With `V = {1}`, `m = r M₁`, `n = r M₀` it spans `I_r` over `J₀` (0-based, so
`I_0 = J₀ + (P)`); with `V = Z(𝔭)` it spans the ideal `A`. -/
noncomputable def obstructionSpan {K A₀ A₁ B₁ : Type*} [Field K] [Fintype B₁]
    (P : MvPolynomial (A₀ ⊕ (A₁ × Bool)) K) (W : Submodule K ((A₀ → K) × (A₁ → K)))
    (u : B₁ → A₀ → K) (a : B₁ → A₁ → Kˣ) (m n : ℕ) (V : Set (TorusGroup K A₀ A₁)) :
    Submodule K (MvPolynomial (A₀ ⊕ (A₁ × Bool)) K) :=
  Submodule.span K {Q | ∃ s : B₁ → ℤ, (∀ j, |s j| ≤ (m : ℤ)) ∧ ∃ v ∈ V,
    ∃ l : List ((A₀ → K) × (A₁ → K)), l.length ≤ n ∧ (∀ w ∈ l, w ∈ W) ∧
      Q = torusTransl (hmePoint u a s * v) (torusDerivList l P)}

/-! Structural lemmas (each a few lines). -/

@[simp] theorem torusDeriv_X_inl {K A₀ A₁ : Type*} [Field K] (w : (A₀ → K) × (A₁ → K)) (h : A₀) :
    torusDeriv w (X (Sum.inl h) : MvPolynomial (A₀ ⊕ (A₁ × Bool)) K) = C (w.1 h) := by
  simp [torusDeriv, mkDerivation_X]

@[simp] theorem torusDeriv_X_inr {K A₀ A₁ : Type*} [Field K] (w : (A₀ → K) × (A₁ → K)) (i : A₁)
    (b : Bool) :
    torusDeriv w (X (Sum.inr (i, b)) : MvPolynomial (A₀ ⊕ (A₁ × Bool)) K)
      = if b then C (w.2 i) * X (Sum.inr (i, b)) else -(C (w.2 i) * X (Sum.inr (i, b))) := by
  simp [torusDeriv, mkDerivation_X]

@[simp] theorem torusDerivList_nil {K A₀ A₁ : Type*} [Field K] :
    torusDerivList ([] : List ((A₀ → K) × (A₁ → K))) = 1 := rfl

theorem mem_torusZero {K A₀ A₁ : Type*} [Field K] {I : Ideal (MvPolynomial (A₀ ⊕ (A₁ × Bool)) K)}
    {g : TorusGroup K A₀ A₁} : g ∈ torusZero I ↔ ∀ f ∈ I, eval (torusPt g) f = 0 := Iff.rfl

theorem eval_torusTransl {K A₀ A₁ : Type*} [Field K] (g h : TorusGroup K A₀ A₁)
    (f : MvPolynomial (A₀ ⊕ (A₁ × Bool)) K) :
    eval (torusPt h) (torusTransl g f) = eval (torusPt (h * g)) f := by
  induction f using MvPolynomial.induction_on with
  | C a => simp [torusTransl]
  | add p q hp hq => simp only [map_add, hp, hq]
  | mul_X p v hp =>
    rw [map_mul, map_mul, hp, map_mul]
    congr 1
    rcases v with i | ⟨i, b⟩
    · simp [torusTransl, torusPt, torusPoint, toAdd_mul, add_comm]
    · cases b <;> simp [torusTransl, torusPt, torusPoint, mul_comm]

end DiazModulus
Source
M. Waldschmidt, Diophantine Approximation on Linear Algebraic Groups, Grundlehren der mathematischen Wissenschaften 326, Springer, 2000, Chapter 8 (by D. Roy): the invariant derivations D_w, Lemma 8.11, and their translation invariance (8.10); the stabiliser, the ideals and the vanishing conditions of the proof of Th. 8.1, §8.4.1. Written here at finite level and in coordinates on the model K[X, Y, Z] of the published Definition DiazHilbertBox. C1.3 of the Diaz modulus mission. Definitions in coordinates: 29 October 2026 (C. Perassi).

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 paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me