OAI/Probability/CoordinateSweeps/Scales.lean

GitHub source file

17 declarations, 0 contain sorry, 0 declared as axioms

Work on the genuinely missing main statements of the pinned manuscript. All definitions are finite and use the source's ordinary probability and trace normalizations. No result of the manuscript is assumed.

  1. lemma OAI.CoordinateSweeps.Grid.exists_concat_of_length line 20

    Source comment

    Any cut retains exactly the ordered bit list; no coordinate permutation is introduced into the sweep law.
    lemma exists_concat_of_length (G : Grid) (n m : ℕ) (hn : 0 < n) (hm : 0 < m)
        (hb : G.b=n+m) : ∃ A B : Grid, A.b=n ∧ B.b=m ∧ concat A B=G := by
    …

    Mentions: OAI.CoordinateSweeps.Grid, OAI.CoordinateSweeps.Grid.concat

  2. lemma OAI.CoordinateSweeps.Grid.exists_half_concat line 38

    Source comment

    Cutting the coordinate list in half is balanced because every bit count is between r and 2r. This avoids choosing a cut by logarithmic rounding.
    lemma exists_half_concat (G : Grid) (hb : 2 ≤ G.b) :
        ∃ A B : Grid, concat A B=G ∧ A.b<G.b ∧ B.b<G.b ∧ A.b≤B.b ∧ B.b≤2*A.b := by
    …

    Mentions: OAI.CoordinateSweeps.Grid, OAI.CoordinateSweeps.Grid.concat

  3. lemma OAI.CoordinateSweeps.Grid.balanced_logs line 51
  4. lemma OAI.CoordinateSweeps.Grid.balanced_sizes line 75
  5. lemma OAI.CoordinateSweeps.ScalarInduction.eventually_log_power_bound line 105

    Source comment

    A fixed polynomial logarithmic loss is uniformly negligible compared with any strictly larger power, the quantitative choice needed in06.
    lemma eventually_log_power_bound (u v r C : ℝ) (huv : u < v) (hC : 0 < C) :
        ∀ᶠ s : ℝ in atTop, 0 < s ∧ s^u*(Real.log s)^r ≤ C*s^v := by
    …
  6. lemma OAI.CoordinateSweeps.ScalarInduction.eventually_constant_le_power line 119
  7. lemma OAI.CoordinateSweeps.ScalarInduction.eventually_absorb_hook line 123
  8. lemma OAI.CoordinateSweeps.ScalarInduction.eventually_absorb_removed line 148
  9. theorem OAI.CoordinateSweeps.ScalarInduction.removed_endgame line 175

    Source comment

    Scalar endgame after the *actual* induced placement-cycle estimate. All representation/path data enter only through their already derived entropies, multiplicity and cost inequalities.
    theorem removed_endgame (L F Fg Fd k h b C : ℝ) (_ : ℝ) (err : ℝ)
        (hL : 0 < L) (_ : 0 ≤ F) (hh : 0 ≤ h) (_ : 0 ≤ k)
        (hD : F ≤ Fg+k*L) (hroot : 0 < Real.sqrt L)
        (hcost : (c0+e0)*k*L+Real.log 4*k*b-Fd+err ≤ F/(20*Real.sqrt L)) :
        -(c0+1/Real.sqrt L+1/(20*Real.sqrt L))*Fg+
          (e0-1/Real.sqrt L-1/(20*Real.sqrt L))*(h+k)*L+err-C+Real.log 4*k*b-Fd ≤
          -(c0+1/Real.sqrt L)*F+(e0-1/Real.sqrt L)*h*L-C := by
    …

    Mentions: OAI.CoordinateSweeps.c0, OAI.CoordinateSweeps.e0

  10. lemma OAI.CoordinateSweeps.ScalarInduction.reciprocal_sqrt_gap line 192

    Source comment

    Child coefficient improvement from a balanced nontrivial split.
    lemma reciprocal_sqrt_gap {L t : ℝ} (hL : 0 < L) (ht : 0 < t) (htL : t ≤ (4/5)*L) :
        (11/10 : ℝ)/Real.sqrt L ≤ 1/Real.sqrt t := by
    …
  11. lemma OAI.CoordinateSweeps.ScalarInduction.hook_comparison line 201
  12. lemma OAI.CoordinateSweeps.ScalarInduction.hook_error_bound line 220
  13. structure OAI.CoordinateSweeps.ScalarInduction.MainSizeReady line 244

    Also declared: ComparatorChallenges/CoordinateSweeps.lean:250 (same declaration text).

  14. lemma OAI.CoordinateSweeps.ScalarInduction.eventually_mainSizeReady line 249
  15. lemma OAI.CoordinateSweeps.ScalarInduction.improved_coefficients line 258
  16. lemma OAI.CoordinateSweeps.ScalarInduction.hook_strengthening line 263
  17. lemma OAI.CoordinateSweeps.ScalarInduction.removed_cost_bound line 288