A five-dimensional space of rational functions of u containing 1, u, u⁻¹ carries a 2×3 configuration iff it is D1, D2a or D2b
ProvedDiazModulus.rational_space_two_by_three_configuration_iffLet be transcendental, that is, (the algebraic numbers); spans and linear independence are over . Let be a -vector subspace of dimension with , that is, for . Suppose every element of lies in , the subfield of generated by and . Then carries a configuration (linearly independent and linearly independent with every ) if and only if one of the following holds:
- (D1) there are linearly independent with and for every ;
- (D2a, two-orbit) for some with coprime, and ;
- (D2b, cyclic) for some such .
Proof. () DiazModulus.rational_space_configuration_coprime_ratio writes with coprime and , and DiazModulus.two_by_three_configuration_ratio_degree_le_two bounds that degree by . Degree : DiazModulus.mobius_ratio_configuration_u_multiplier_triple gives (D1). Degree : Theorem NF (DiazModulus.degree_two_configuration_normal_form) gives (D1), (D2a) or (D2b). () Under (D1), is a configuration, being independent as . Under (D2a) or (D2b), is the canonical span, of dimension , and DiazModulus.quadratic_two_orbit_space_carries_two_by_three_configuration, resp. DiazModulus.quadratic_cyclic_space_carries_two_by_three_configuration, give the configuration .
Novelty. Not found in the sources read: the “only if” half, the classification. The configurations themselves (the “if” half) give, at a candidate, exclusions that are substitutions in Roy's strong six exponentials theorem (Roy 1992, Cor. 2). Applied to the conjugation hulls of rational , it gives the five-dimensional case of Theorem F (DiazModulus.rational_conj_hull_two_by_three_configuration_iff). Theorem F asks the question of Theorem E of the mission's research note (the Laurent case, formal as DiazModulus.laurent_conj_hull_two_by_three_configuration_iff) for every rational in , and builds on Theorem B (DiazModulus.circle_point_extension_two_by_three_configuration_iff, 10.6), which holds for every complex . For Laurent , that its answer gives back Theorem E's classes (P), (Q) and (R) was checked by exact computation on random data, not proved; Theorem E's explicit criteria do not follow from it.
import Definitions.Def_DiazModulus
open Complex ComplexConjugate
namespace DiazModulus
open Polynomial in
theorem rational_space_two_by_three_configuration_iff (u : ℂ) (hu : u ∉ Qbar) (W : Submodule (↥Qbar) ℂ) [FiniteDimensional (↥Qbar) W]
(hW : Module.finrank (↥Qbar) W = 5) (h1 : (1 : ℂ) ∈ W) (hu1 : u ∈ W) (hui : u⁻¹ ∈ W)
(hrat : ∀ w ∈ W, w ∈ IntermediateField.adjoin Qbar ({u} : Set ℂ)) :
(∃ (x : Fin 2 → ℂ) (y : Fin 3 → ℂ), LinearIndependent (↥Qbar) x ∧ LinearIndependent (↥Qbar) y ∧
∀ i j, x i * y j ∈ W) ↔
(∃ t : Fin 3 → ℂ, LinearIndependent (↥Qbar) t ∧ ∀ j, t j ∈ W ∧ u * t j ∈ W) ∨ (∃ P' Q' S₁ S₂ : (↥Qbar)[X], IsCoprime P' Q' ∧ max P'.natDegree Q'.natDegree = 2 ∧
S₁.natDegree ≤ 2 ∧ S₂.natDegree ≤ 2 ∧
W = Submodule.span Qbar ({1, u, u⁻¹, aeval u Q' * aeval u S₁ / (u * aeval u P'),
aeval u P' * aeval u S₂ / (u * aeval u Q')} : Set ℂ)) ∨
(∃ P' Q' S₁ S₂ : (↥Qbar)[X], IsCoprime P' Q' ∧ max P'.natDegree Q'.natDegree = 2 ∧
S₁.natDegree ≤ 2 ∧ S₂.natDegree ≤ 2 ∧
W = Submodule.span Qbar ({1, u, u⁻¹, aeval u Q' * aeval u S₁ / (u * aeval u P'),
aeval u Q' * (aeval u Q' * aeval u S₁ + aeval u P' * aeval u S₂) / (u * aeval u P' ^ 2)} :
Set ℂ)) := by
sorry
end DiazModulus