Bulk recovery at the level of exponents.
ProvedHarmonicBulkSteeperEdge.localExp_mix_tendstoaether-catalogprobability
Bulk recovery at the level of exponents. The local exponent of a bulk × edge mixture converges to the bulk exponent.
theorem HarmonicBulkSteeperEdge.localExp_mix_tendsto{w a b : ℝ} (hw0 : 0 < w) (hw1 : w < 1) (hab : a < b) :
Tendsto (fun k : ℕ => localExp (mix w a b) k) atTop (nhds a) := by sorry
Formalization Note Transplanted verbatim from the Aether Catalog source Probability/HarmonicBulkSteeperEdge.lean; the statement is byte-identical to the source declaration, elaborated with autoImplicit disabled in the platform environment.
Preamble
-- Thm stub generated from Probability/HarmonicBulkSteeperEdge.lean
import Mathlib
import Definitions.Def_Probability_HarmonicBulkSteeperEdge
/-
Copyright (c) 2026 Harmonic. All rights reserved.
Released under Apache 2.0 license.
# Harmonic bulk × steeper edge: rigidity and mixture resolution for discrete power-law kernels
This file formalises the "harmonic bulk × steeper-edge kernel" refinement question:
a fitted bulk exponent (near the harmonic value `a = 1`, empirically `1.104`) coexists
with head statistics (edge fraction, first-decile mass, peak/end ratio) that each imply a
*steeper* exponent. The mathematical content developed here is:
* **MLR ⇒ FOSD for the discrete power-law family.** For weights `k ↦ k^(-a)` on
`{1, …, n}`, a larger exponent puts more mass on every head window
(`headMass_le_of_exponent_le`), strictly so (`headMass_lt_of_exponent_lt`).
* **Rigidity / identifiability.** A single head statistic determines the exponent
uniquely (`exponent_unique_of_headMass_eq`); hence two head windows reporting
*different* implied exponents cannot be reconciled by any single power law
(`no_single_exponent_fits_two_windows`). This is the exact logical shape of the
recorded tension.
* **Two-component resolution.** For a mixture of a flat (bulk) and a steep (edge)
component, every local exponent lies strictly between the two component exponents
(`localExp_mix_mem_Ioo`), the head mass is a strict mediant
(`headMass_mix_mem_Ioo`), and the steep component's local share is strictly
decreasing in the index (`steepShare_strictAnti`) with limit `0`
(`steepShare_tendsto_zero`). So the steep component is *exactly* an edge phenomenon
while the bulk is governed by the flat exponent.
* **A quantitative instance of the tension and its resolution.** No power law with
exponent `≤ 1.104` can produce the recorded peak/end ratio `2.54`
(`pure_power_law_peak_end_lt_observed`), while the explicit harmonic-bulk /
quadratic-edge mixture with weight `w = 54/127` produces it exactly
(`harmonic_edge_mixture_matches_observed_peak_end`).
-/
open Finset Filter
open HarmonicBulkSteeperEdge
/-! ## The discrete power-law kernel -/
/-! ## Head masses and first-order stochastic dominance -/
/-! ## Mediants -/
/-! ## Two-component (bulk × edge) kernels -/
/-! ## The steep component is an edge phenomenon -/
/-! ## A quantitative instance: harmonic bulk versus the recorded peak/end ratio -/
/-! ## Quantitative bulk recovery for the local exponent
The local exponent of a bulk × edge mixture exceeds the bulk exponent at every index, but
only by `O(k^{-(b-a)})`: the edge steepening is a genuinely local excess that decays at a
power rate, and the measured exponent converges to the bulk exponent. -/Formal statement
theorem HarmonicBulkSteeperEdge.localExp_mix_tendsto{w a b : ℝ} (hw0 : 0 < w) (hw1 : w < 1) (hab : a < b) :
Tendsto (fun k : ℕ => localExp (mix w a b) k) atTop (nhds a) := by sorrySource