Bounded modular certificates characterize rational image dimension
ProvedAPSCounting893.Main.bounded_modular_dimension_certificatesLet be a valid arithmetic circuit over , including circuits with division gates whose outputs are promised polynomials. Let be its polynomial map over , and let be the Zariski closure of its image. Let be the bounded modular Jacobian-certificate predicate defined in the accompanying module. Then there is a natural number such that
In this predicate, a certificate consists of distinct output and variable indices, a prime , and a sparse integer point where the cleared Jacobian minor is nonzero modulo . For circuit encoding length , selected variable indices have at most bits, has at most bits, and the encoded sparse point has length at most .
This connects algebraic image dimension with bounded finite arithmetic data. It applies even when the declared number of variables is given in binary and many variables are unused. The empty-output and zero-dimensional cases are included. This theorem supplies the mathematical certificate equivalence; the uniform finite-machine search is a separate theorem.
import Definitions.Def_APSCounting893_Main_ModularJacobianData set_option autoImplicit false
theorem APSCounting893.Main.bounded_modular_dimension_certificates {n : ℕ} (C : APSCounting893.Main.ArithmeticCircuit n)
(hvalid : APSCounting893.Main.ValidCircuit .rational C) :
∃ r : ℕ,RSCapacity.algDim (APSCounting893.Main.imageClosure (APSCounting893.Main.rationalComplexPolynomials C))=
((r : ℕ∞) : WithBot ℕ∞) ∧
∀ k : ℕ,k≤r ↔ APSCounting893.Main.ModularJacobianCertificate C k := by sorry