Theorem 3.2 — Upper unitriangular Wronskian factorization
OpenDerivationHitting.Commutative.Supplement.wronskian_factorizationLet F be any characteristic-zero field and let f∈F[x₁,…,xₙ]. Let g₁,…,gᵣ be a degree-filtration-compatible basis of the whole ordinary derivative space of f, ordered by total degree. For any rational kernel tuple bᵢ(t), form D=∂ₜ+Σᵢbᵢ(t)∂ₓᵢ and the matrix W_D whose row m is (Dᵐgⱼ)ⱼ, for 0≤m<r. Then there exist matrices R over F(t) and H over F[x₁,…,xₙ] such that
The identity is in the matrix ring over F(t)[x]. Both triangularity and every diagonal entry are asserted; the polynomial factor has no dependence on t. The rational kernels are unrestricted. Nonzero constants require no normalization, and r=0 gives the empty matrix identity.
This companion theorem asserts a matrix factorization, in addition to the determinant statement already published. The basis reading follows the source's earlier-index derivative relations; a merely degree-sorted basis can fail, as documented with the rational one-variable cubic counterexample in the moderation notes.
Differences from the source:
- encoding: variables numbered 1,…,n → variables Fin n with source index i.val+1; reason: a bijective change of indices preserves every coefficient
- reading: §3 chooses a degree-ordered basis and uses strictly earlier-index derivative relations → the chosen basis is compatible with every degree-filtration piece as well as ordered by degree; reason: mere sorting is insufficient; the source argument requires filtration compatibility, which is available for every derivative space
- generalization: the setup normalizes the first basis vector to 1 and the last to f → any degree-compatible basis, including unnormalized bases for nonzero constants and the empty basis for f=0; reason: the factorization is invariant under these scalar basis normalizations and specializes to the source setup
- encoding: W_D=R(t)F(x) → both factors are embedded into F(t)[x] before matrix multiplication; reason: the types enforce that R has no x dependence and the polynomial factor has no t dependence
import Definitions.Def_DerivationHitting_Commutative_Supplement_WronskianData
namespace DerivationHitting.Commutative.Supplement
theorem wronskian_factorization {F : Type*} [Field F] [CharZero F] {n r : ℕ}
(f : MvPolynomial (Fin n) F) (g : Fin r → MvPolynomial (Fin n) F)
(hg : DegreeCompatibleBasis f g) (b : Fin n → RatFunc F) :
∃ (R : Matrix (Fin r) (Fin r) (RatFunc F))
(H : Matrix (Fin r) (Fin r) (MvPolynomial (Fin n) F)),
derivationWronskianMatrix b g =
R.map MvPolynomial.C * H.map rationalPolynomial ∧
(∀ i j, j < i → H i j = 0) ∧ (∀ i, H i i = 1) := by sorry
end DerivationHitting.Commutative.Supplement