Rational complex matrix certificates preserve strict quadratic-form bounds
ProvedMasulloRH.finite_rational_complex_gram_certificateapproximationcomplex-numberslinear-algebrariemann-hypothesis
Let be a finite index set, a coefficient vector, an arbitrary complex matrix, and . Define
If , then there are rational pairs and a rational number such that, writing ,
This converts a strict bound on an exact finite quadratic form into a finite rational matrix approximation with an explicit allowance for entrywise error. No Hermitian symmetry or positivity assumption on is required. When the real and imaginary parts of and the threshold are rational, the displayed certificate inequality involves only rational quantities.
Formalization Note. Rational complex entries are represented by pairs of rationals. The theorem applies to every finite type, including the empty type.
Preamble
import Mathlib.Analysis.Complex.Norm import Mathlib.Algebra.Order.Archimedean.Real.Basic import Mathlib.Tactic.Linarith import Mathlib.Tactic.Ring import Mathlib.Tactic.Positivity set_option autoImplicit false open scoped BigOperators ComplexConjugate
Formal statement
theorem MasulloRH.finite_rational_complex_gram_certificate {ι : Type*} [Fintype ι]
(d : ι → ℂ) (G : ι → ι → ℂ) (b : ℝ)
(hgap : (∑ i, ∑ j, conj (d i) * d j * G i j).re < b) :
∃ (R : ι → ι → ℚ × ℚ) (η : ℚ), 0 ≤ η ∧
(∀ i j, ‖G i j - (((R i j).1 : ℂ) + ((R i j).2 : ℂ) * Complex.I)‖ ≤ (η : ℝ)) ∧
(∑ i, ∑ j, conj (d i) * d j *
(((R i j).1 : ℂ) + ((R i j).2 : ℂ) * Complex.I)).re +
(η : ℝ) * (∑ i, (|(d i).re| + |(d i).im|)) ^ 2 < b := by sorrySource
Elementary finite-dimensional rounding lemma formalized for this contribution, supporting MasulloRH.cofinal_rational_theta_gram_interval_tables, https://prove2.me/theorems/7f59c68d-29e6-404c-a754-27288c412b6c . The source target fixes the entrywise-error certificate and coefficient bound used here. This is original supporting formalization, not a theorem claimed to be extracted from Masullo, The Music of the Primes. Its proof uses rational density and the triangle inequality.