TaoFivePrimes.rh_high_height_vector_fft_semantic
ProvedTaoFivePrimes.rh_high_height_vector_fft_semanticnumerical-verificationriemann-zetatao-five-primes
The ordinary integer packed convolution output approximates the explicit discrete Fourier transform of any proposed sparse vector bounded by18, with error at most10^(-13). All packing, convolution, rounding and chirp errors are proved. Makes no analytic polynomial, sign, count or RH assertion.
Preamble
import Mathlib import Definitions.Def_TaoFivePrimes_RHHighContourVector import Definitions.Def_TaoFivePrimes_RHHighFFTArithmeticBase import Definitions.Def_TaoFivePrimes_RHHighPackedFFT
Formal statement
theorem TaoFivePrimes.rh_high_height_vector_fft_semantic (v : List Nat) (ar ai : Fin 16384→Nat) (br bi : Fin 32767→Nat)
(hAr:∀ j,ar j≤2^167) (hAi:∀ j,ai j≤2^167)
(hBr:∀ j,br j≤2^82) (hBi:∀ j,bi j≤2^82)
(hmag:∀j:Fin 16384,‖RHHighContourVector.value (RHHighContourVector.vector v j.val)‖≤18)
(hinput:∀ j:Fin 16384,RHHighContourVector.raw v RHHighFFTArithmeticBase.chirp j.val=
((ar j:Int)-2^166,(ai j:Int)-2^166))
(hkernel:∀ j:Fin 32767,
((br j:Int)-2^81,(bi j:Int)-2^81)=
RHHighFFTArithmeticBase.chirp ((j.val:Int)-16383).natAbs)
(r : Fin 2048) :
‖(∑j:Fin 16384,RHHighContourVector.value (RHHighContourVector.vector v j.val)*
Complex.exp (((Real.pi*(-2*(j.val:ℝ)*(r.val:ℝ)/16384):ℝ):ℂ)*Complex.I))-
(let p:=(((RHHighPackedFFT.output (List.ofFn ar) (List.ofFn ai)
(List.ofFn br) (List.ofFn bi) (List.ofFn (fun j=>br j+bi j))
RHHighFFTArithmeticBase.chirp)[r.val]?).getD (0,0))
((p.1:ℂ)+(p.2:ℂ)*Complex.I)/(2:ℂ)^80)‖≤1/10^13 := by sorrySource
Existing Platt Fourier construction, DOI10.1090/mcom/3198. Original surrounding attribution sometik179 c9908072-d99a-4431-aad8-a72c6352619a and Gabewhigham.