Documentation

LeanPool.Superdiffusion.Algsuperdiff.Section4.Provider.Proportion.G2ThresholdSharp

G2ThresholdSharp #

Ported from scottnarmstrong/Superdiffusion at 7f20476385d4af418f110303085ad0740c9eba46. Module imports, support namespace and local option directives adapted for Lean Pool.

The 𝒢₂ proportion endpoint at the θ-free ε-threshold #

G2Threshold.ratioTail_eventG2_of_threshold_shift restates the proved lane at the named threshold g2Threshold … = 36(1+C⋆)D/θ. This module restates it at the sharp threshold g2ThresholdSharp … = 108(1+C⋆)D, which carries no divisor θ: the quantile factor θ^{-1/p} is paid logarithmically (G2SharpArith.threshold_le_inv_sharp) rather than by θ^{-1/p} ≤ θ^{-1}, and the lane's own moment exponent already clears the requirement θ⁻¹ ≤ p (G2SharpArith.inv_theta_le_g2MomentExponent).

Nothing else in the lane changes: the moment exponent, the normalizer, the independence geometry, the per-cube display and the reduction are the proved ones, character for character.

Contents #

The headline #

At the anchor's own clauses the sharp threshold condition reduces to the two parameter-free constant floors

g2SharpAnchorConst d r C N ≤ C_a      and      g2SharpAnchorConstTwo d r C N ≤ C_a ,

both functions of d alone. There is no new anchor clause: in particular nothing of the shape C_aθ ≥ κ(d)s that the proved G2Threshold.g2Threshold_le_of_anchor needed. The anchor's own level clause C_a c⋆^{-2}s^{-6}ε^{-2}γ ≤ θ does all the work, because the sharp threshold's binding clause carries only one power of θ on its favourable side, and the level clause supplies exactly one.

Scope #

Provider material: proved local helpers. The ε-threshold condition of the endpoint is a genuine hypothesis of the lane, not a source premise smuggled into a frozen statement, and the two constant floors are explicit named parameter conditions.

References #

1. The endpoint at the θ-free threshold #

theorem Algsuperdiff.Section4.Provider.Proportion.ratioTail_eventG2_of_thresholdSharp_shift (d : ℕ) :
∃ (C : ℝ), 0 < C ∧ ∀ (r : ℕ), 1 ≤ r → 3 + 2 * 3 ^ (1 - 2) * √↑d ≤ 3 ^ r → ∀ (M : Section3.ABKModel d), M.gamma ≤ C⁻¹ ^ 10 * Section3.Disorder.cstar M ^ 10 → ∀ (s : { s : ℝ // 0 < s }), ↑s ∈ Set.Icc (8 * M.gamma) 1 → ∀ (theta : ℝ), 0 < theta → theta * (↑r + 1) < 1 → ∀ (c1 : ℝ), 0 ≤ c1 → ∀ (ep : ℝ), 0 < ep → g2ThresholdSharp d r C M (↑s) theta c1 ≤ ep ^ 2 → ∀ (m0 : ℤ) (n : ℕ), ↑(Section3.Cutoff.cutoffSampleLaw M) {omega : Section3.Cutoff.CutoffSample d | theta < Probability.IndicatorDensity.scaleProp (fun (k : ℤ) => (Support.eventG2 M (m0 + k) s ep)ᶜ) n omega} ≤ ENNReal.ofReal (Real.exp (-c1 * ↑n) / 3)

This is the shifted 𝒢₂ lane endpoint with its existential threshold ∃ ε₀ > 0, ∀ ε ≥ ε₀ replaced by the named condition

0 < ε   and   g2ThresholdSharp d r C M s θ c₁ ≤ ε² ,

g2ThresholdSharp … = 108(1+C⋆)·D being a closed-form expression in the lane's parameters with no divisor θ (contrast g2Threshold … = 36(1+C⋆)D/θ). Every other hypothesis is that of the proved endpoint, unchanged: the honest range condition, the printed regime γ ≤ C^{−10}c⋆^{10}, the standing window s ∈ [8γ,1], the level condition θ(r+1) < 1 and c₁ ≥ 0. The base m₀ and the window n are quantified last, so the conclusion is the LaneTailFrom shape the proportion assembly consumes.

2. The sharp threshold at the frozen theorem's own parameters #

Unlike its proved counterpart G2Threshold.g2AnchorConst, this constant is asked to clear C_a outright: the anchor-side condition is g2SharpAnchorConst d r ≤ C_a, with no s and no θ in it.

Equations
Instances For

    The constant-size floor the anchor's own constant must clear so that the exponentially small clause of the sharp threshold condition is free. Like g2SharpAnchorConst it constrains only C_a against d.

    Equations
    Instances For
      theorem Algsuperdiff.Section4.Provider.Proportion.g2ThresholdSharp_le_of_anchor_rate {d : ℕ} (M : Section3.ABKModel d) {C Ca N s ep theta c1 : ℝ} {r : ℕ} (hr1 : 1 ≤ r) (hC0 : 0 < C) (hN1 : 1 ≤ N) (hCa1 : 1 ≤ Ca) (hs0 : 0 < s) (hs12 : s ≤ 1 / 2) (hep0 : 0 < ep) (hep12 : ep ≤ 1 / 2) (hth0 : 0 < theta) (hth12 : theta ≤ 1 / 2) (hc10 : 0 ≤ c1) (hreg : M.gamma ≤ Ca⁻¹ * Section3.Disorder.cstar M ^ 10) (hlev : M.gamma ≤ theta * Section3.Disorder.cstar M ^ 2 * s ^ 6 * ep ^ 2 / Ca) (hZ : M.gamma * (1 + c1) ≤ 2 * (theta * Section3.Disorder.cstar M ^ 2 * s ^ 6 * ep ^ 2 / Ca)) (hkappa : g2SharpAnchorConst d r C N ≤ Ca) (hCaBig : g2SharpAnchorConstTwo d r C N ≤ Ca) :
      g2ThresholdSharp d r C M s (theta / N) c1 ≤ ep ^ 2

      The sharp threshold condition at the anchor's clauses, for an arbitrary rate.

      The rate enters only through hZ, the requirement that γ(1+c₁) still obeys the anchor's level clause up to a factor 2. Everything else is the anchor's own data: its parameter ranges s, ε, θ ∈ (0,½], its regime clause γ ≤ C_a^{-1}c⋆^{10}, its s⁻⁶ level clause in solved form hlev, and the two parameter-free constant floors hkappa, hCaBig.

      This is the item-9 payoff: compare G2Threshold.g2Threshold_le_of_anchor_rate, whose corresponding hypothesis is g2AnchorConst d r · s ≤ C_a · θ — a genuine new clause coupling θ to s, which the verified consumer refutes. Here there is no such clause at all.

      theorem Algsuperdiff.Section4.Provider.Proportion.g2ThresholdSharp_le_of_anchor_solved {d : ℕ} (M : Section3.ABKModel d) {C Ca N s ep theta : ℝ} {r : ℕ} (hr1 : 1 ≤ r) (hC0 : 0 < C) (hN1 : 1 ≤ N) (hCa1 : 1 ≤ Ca) (hs0 : 0 < s) (hs12 : s ≤ 1 / 2) (hep0 : 0 < ep) (hep12 : ep ≤ 1 / 2) (hth0 : 0 < theta) (hth12 : theta ≤ 1 / 2) (hreg : M.gamma ≤ Ca⁻¹ * Section3.Disorder.cstar M ^ 10) (hlev : M.gamma ≤ theta * Section3.Disorder.cstar M ^ 2 * s ^ 6 * ep ^ 2 / Ca) (hkappa : g2SharpAnchorConst d r C N ≤ Ca) (hCaBig : g2SharpAnchorConstTwo d r C N ≤ Ca) :
      g2ThresholdSharp d r C M s (theta / N) (2 * (Section3.Disorder.cstar M ^ 2 * s ^ 7 * ep ^ 2 * theta / (Ca * M.gamma))) ≤ ep ^ 2

      The sharp threshold condition at the assembly rate c₁ = 2A, from the level clause in solved form.

      A = c⋆²s⁷ε²θ/(C_aγ) is the rate the anchor clause carries. The grouping γ(1+c₁) = γ + 2sX, X = θc⋆²s⁶ε²/C_a, linearises the requirement at once: γ ≤ X and 2s ≤ 1 give γ(1+c₁) ≤ 2X.

      theorem Algsuperdiff.Section4.Provider.Proportion.g2ThresholdSharp_le_of_anchor {d : ℕ} (M : Section3.ABKModel d) {C Ca N s ep theta : ℝ} {r : ℕ} (hr1 : 1 ≤ r) (hC0 : 0 < C) (hN1 : 1 ≤ N) (hCa1 : 1 ≤ Ca) (hs0 : 0 < s) (hs12 : s ≤ 1 / 2) (hep0 : 0 < ep) (hep12 : ep ≤ 1 / 2) (hth0 : 0 < theta) (hth12 : theta ≤ 1 / 2) (hreg : M.gamma ≤ Ca⁻¹ * Section3.Disorder.cstar M ^ 10) (hlevel : Ca * (Section3.Disorder.cstar M)⁻¹ ^ 2 * s⁻¹ ^ 6 * ep⁻¹ ^ 2 * M.gamma ≤ theta) (hkappa : g2SharpAnchorConst d r C N ≤ Ca) (hCaBig : g2SharpAnchorConstTwo d r C N ≤ Ca) :
      g2ThresholdSharp d r C M s (theta / N) (2 * (Section3.Disorder.cstar M ^ 2 * s ^ 7 * ep ^ 2 * theta / (Ca * M.gamma))) ≤ ep ^ 2

      The lane is read at level θ/N and at the assembly rate c₁ = 2A, A = c⋆²s⁷ε²θ/(C_aγ). Under the anchor's parameter ranges, its regime clause and its s⁻⁶ level clause, the whole ε-threshold condition reduces to the two parameter-free constant floors

      g2SharpAnchorConst d r C N ≤ C_a      and      g2SharpAnchorConstTwo d r C N ≤ C_a ,
      

      both functions of d alone (once the lane's range r(d), its constant C(d) and the level divisor N are fixed). There is no residual clause in θ or s.

      This resolves the 𝒢₂ side: the proved G2Threshold.g2Threshold_le_of_anchor needed g2AnchorConst d r · s ≤ C_aθ, which the verified consumer refutes as θ → 0; the log-quantile reading removes it entirely.