Finite dyadic theta sums with a rational-cutoff reflected evaluation
DefinitionMasulloRH_GaussianThetaTruncationapproximationgaussianmasullorigorous-numericstheta-functions
For and , define
Also define the reflected finite approximation
These are finite expressions in real elementary functions, embedded in . The rational switch at allows shifted copies to be integrated on subintervals with rational breakpoints. The definitions contain no accuracy claim; comparison with the actual theta series and its reflection is proved separately. The finite reflected approximation can have a small jump at the cutoff, so integration must respect its piecewise definition.
Formalization Note. The Lean names are MasulloRH.GaussianApproximation.thetaPartial and MasulloRH.GaussianApproximation.thetaPartialReflected. The displayed positive-index sum is represented by indices with argument .
Definition code
import Mathlib.Analysis.Complex.Exponential
import Mathlib.Analysis.SpecialFunctions.Log.Basic
import Mathlib.Analysis.SpecialFunctions.Trigonometric.Basic
set_option autoImplicit false
open scoped BigOperators
namespace MasulloRH.GaussianApproximation
noncomputable section
/-- The finite symmetric truncation of the mission's dyadic Gaussian theta series. -/
def thetaPartial (K : ℕ) (x : ℝ) : ℂ :=
2 * ∑ n ∈ Finset.range K,
((Real.exp (-Real.pi * (((n : ℝ) + 1) * Real.exp x) ^ 2) -
3 * Real.exp (-Real.pi * (2 * (((n : ℝ) + 1) * Real.exp x)) ^ 2) +
2 * Real.exp (-Real.pi * (4 * (((n : ℝ) + 1) * Real.exp x)) ^ 2) : ℝ) : ℂ)
/-- Evaluate on the reflected side below the rational cutoff -1. The rational
cutoff lets numerical quadrature split shifted copies at rational points. -/
def thetaPartialReflected (K : ℕ) (x : ℝ) : ℂ :=
if -1 ≤ x then thetaPartial K x
else ((Real.exp (-x) / 2 : ℝ) : ℂ) * thetaPartial K (-x - Real.log 4)
end
end MasulloRH.GaussianApproximation
Source
Finite sum from the proved theta Gaussian truncation theorem, https://prove2.me/theorems/2e9c99f7-cd06-43f7-a85b-f4d2cc422151 , for the actual mission definition https://prove2.me/definitions/6f34c003-4870-4c6a-b83e-256b289a15db . The reflected branch is chosen using the exact Poisson identity theta(x)=exp(-x)/2*theta(-x-log4), with a rational cutoff for numerical integration. Supporting definitions introduced for this contribution; no additional axioms or accuracy premises.