Lemma 2.1 — power-sum ratio bound
ProvedWangRenyiEntropy.Noninteger.power_sum_ratiop2o-batch-stoc27ap2o-gran-per-chapterp2o-plan-paperp2o-v1renyi-entropystatistical-estimation
Let , be any real order, including positive integers, and let be any probability law on an alphabet of size . Then
Both power sums have positive order and the denominator is positive by normalization. The bound controls the relative-variance contribution without excluding zero coordinates or singleton alphabets.
Differences from the source:
- encoding: finite alphabet and nonnegative normalized real probability coordinates → Fin d with d ≥ 1 and the published IsLaw predicate, allowing zero coordinates; reason: relabeling preserves every finite nonempty alphabet and every probability law
Preamble
import Mathlib import Definitions.Def_WangRenyiEntropy_Noninteger_SamplingExperiment
Formal statement
namespace WangRenyiEntropy.Noninteger
/-- Lemma 2.1, printed p.6 / PDF p.8. No noninteger restriction. -/
theorem power_sum_ratio (α : ℝ) (hα : 1 / 2 < α) (hα1 : α ≠ 1)
(d : ℕ) (hd : 1 ≤ d) (p : Law d) :
powerSum (2 * α - 1) p / (powerSum α p) ^ 2 ≤ Real.rpow d |1 - 1 / α| := by sorry
end WangRenyiEntropy.Noninteger
Source
Qisheng Wang, STOC 2027 submission #447 (October 2026), https://github.com/sigact-stoc/stoc2027/releases/download/papers-1/447.pdf, printed 6 / PDF 8, Lemma 2.1
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.