Integer dyadic scaling sharpens logarithm enclosures
Provedmme_dyadic_log_series_boundsmatrix-multiplicationtensor-extraction
Integer powers of two move small probabilities and large dual weights near one. Finite series for the scaled argument and log two give explicit rational centers and errors for the original logarithm. The root exponent bound remains a separate obligation.
Preamble
import Theorems.Thm_mme_log_series_bounds open scoped BigOperators
Formal statement
theorem mme_dyadic_log_series_bounds (x : ℝ) (hx : 0 < x) (k : ℤ) (n : ℕ) :
let t := ((2 : ℝ) ^ k * x - 1) / ((2 : ℝ) ^ k * x + 1)
let twoCenter := 2 * ∑ i ∈ Finset.range n, (1 / 3 : ℝ) ^ (2 * i + 1) / (2 * i + 1)
let twoError := 2 * ((1 / 3 : ℝ) ^ (2 * n + 1) / (1 - (1 / 3 : ℝ) ^ 2))
let center := (2 * ∑ i ∈ Finset.range n, t ^ (2 * i + 1) / (2 * i + 1)) -
(k : ℝ) * twoCenter
let error := 2 * (|t| ^ (2 * n + 1) / (1 - t ^ 2)) + |(k : ℝ)| * twoError
center - error ≤ Real.log x ∧ Real.log x ≤ center + error := by sorrySource
Checked tensor restrictions and released global histogram extraction.