Strict two-sided noise.
ProvedEnergyAscent.channel_strictly_noisyaether-catalogcombinatorics
Strict two-sided noise. At every scale the middle band contains both a window hit and a window miss, so the magnitude channel and the tree letter determine each other in exactly one direction: the value of the channel is bounded away from a perfect oracle.
theorem EnergyAscent.channel_strictly_noisy(S : ℤ) :
(∃ T : ℤ × ℤ × ℤ, S < T.2.1 ∧ Admissible 1 T ∧ InWindow 1 T ∧ LetterOne T) ∧
(∃ T : ℤ × ℤ × ℤ, S < T.2.1 ∧ Admissible 1 T ∧ ¬ InWindow 1 T ∧ LetterOne T) := by sorry
Formalization Note Transplanted verbatim from the Aether Catalog source Combinatorics/EnergyAscentChannel.lean; the statement is byte-identical to the source declaration, elaborated with autoImplicit disabled in the platform environment.
Preamble
-- Thm stub generated from Combinatorics/EnergyAscentChannel.lean import Mathlib import Definitions.Def_Combinatorics_EnergyAscentChannel import Definitions.Def_Combinatorics_EnergyAscentPellSpine /-! # Energy-Ascent V: the magnitude channel, its positive dependence and its ceiling Energy-Ascent III showed that above scale `112·W` a Fermat-window hit forces the middle ratio band, and Energy-Ascent IV showed that hits occur at every scale. Here we draw the two information-theoretic consequences that the experimental round reported, in exact combinatorial form. * **Positive dependence** (the measured `MI = 0.1836 bits`, `z ≈ +110`): on *any* finite family of admissible triples that contains at least one hit and at least one triple outside the middle band, the empirical conditional frequency of letter `1` given a hit *strictly exceeds* its marginal frequency — `EnergyAscent.empirical_positive_dependence`. Equivalently the empirical mutual information of the pair (hit bit, letter) is strictly positive. * **A ceiling on the value** (the reported "value bounded, ~19%"): the channel is strictly noisy in the other direction. We construct an explicit primitive family `noisy k` sitting in the middle band at every scale which the window never sees — `EnergyAscent.noisy_not_hit`. So the letter does not determine the hit bit, and the window bit cannot be upgraded to an equivalence. Both statements are unconditional theorems, not sample statistics. -/ open EnergyAscent open scoped Classical /-! ## The ceiling: an unseeable middle-band family -/
Formal statement
theorem EnergyAscent.channel_strictly_noisy(S : ℤ) :
(∃ T : ℤ × ℤ × ℤ, S < T.2.1 ∧ Admissible 1 T ∧ InWindow 1 T ∧ LetterOne T) ∧
(∃ T : ℤ × ℤ × ℤ, S < T.2.1 ∧ Admissible 1 T ∧ ¬ InWindow 1 T ∧ LetterOne T) := by sorrySource