Explicit error for finite reflected theta evaluation with a rational cutoff
ProvedMasulloRH.theta_reflected_truncation_errorLet be the mission's actual dyadic Gaussian theta series. Let be its symmetric finite sum over the nonzero integers from to , and use the published rational-cutoff reflected evaluation
For every natural and real ,
Thus, for , the error of a shifted evaluation at is at most . The estimate supplies an explicit finite-evaluation allowance on any compact integration window. It has no RH or Mertens assumption and includes .
Formalization Note. The finite functions are defined in MasulloRH_GaussianThetaTruncation. The formal conclusion is the pointwise estimate; the displayed compact-window consequence follows by monotonicity of the exponential. The finite reflected expression may jump at its cutoff, so numerical integration should split there.
import Definitions.Def_MasulloRH_GaussianThetaApproximation import Definitions.Def_MasulloRH_GaussianThetaTruncation open MasulloRH.GaussianApproximation set_option autoImplicit false
theorem MasulloRH.theta_reflected_truncation_error (K : ℕ) (x : ℝ) :
‖theta x - thetaPartialReflected K x‖ ≤
(24 * Real.exp (-((K : ℝ) + 1) ^ 2 / 3)) *
max 1 (Real.exp (-x) / 2) := by sorry