Simultaneous bivariate specialization at a bounded integer
ProvedTranscendenceTheory.bounded_bivariate_specializationLet R be an integral domain of characteristic zero, let I be a finite index type, and let F be a polynomial in R[T][Y]. Let a be a natural number such that every coefficient of F, viewed as a polynomial in Y, has degree in T at most a. Let c_i be any family of elements of R, and let P(i) be any predicate specifying which indices are active.
The following are equivalent:
- For every active i, the polynomial F(T,c_i) is nonzero in R[T].
- There exists a nonnegative integer n with
such that, for every active i,
The same n works for all active indices. In the formal statement, n is an element of Fin(|I|a+1), and F(n,Y) is obtained by mapping each coefficient of F through evaluation at the image of n in R.
Repeated c_i, empty index types, an empty active set, and a=0 are included. There is no positivity assumption on the degree of F in Y. No field or algebraic-closedness hypothesis is required. Characteristic zero ensures that the integer sampling points are distinct.
This is a quantitative simultaneous-specialization equivalence derived from the polynomial root bound. It asserts no preservation of the degree in Y after specialization.
import Mathlib.Algebra.Polynomial.Roots import Mathlib.Algebra.Polynomial.Bivariate import Mathlib.Algebra.Polynomial.Eval.Degree
theorem TranscendenceTheory.bounded_bivariate_specialization
(R : Type*) [CommRing R] [IsDomain R] [CharZero R]
(ι : Type*) [Fintype ι] (F : Polynomial (Polynomial R))
(a : ℕ) (hdegree : ∀ j, (F.coeff j).natDegree ≤ a)
(c : ι → R) (active : ι → Prop) :
(∀ i, active i → F.eval (Polynomial.C (c i)) ≠ 0) ↔
∃ n : Fin (Fintype.card ι * a + 1), ∀ i, active i →
(F.map (Polynomial.evalRingHom (n.val : R))).eval (c i) ≠ 0 := by sorry